PowerHA SystemMirror For IBM I Cookbook
PowerHA SystemMirror For IBM I Cookbook
PowerHA SystemMirror
for IBM i Cookbook
Hernando Bedoya
Abdell Ali-Darwish
Ingo Dimmer
Sabine Jordan
KyoSeok Kim
Akinori Mogi
Nandoo Neerukonda
Tomasz Piela
Marc Rauzier
ibm.com/redbooks
International Technical Support Organization
January 2012
SG24-7994-00
Note: Before using this information and the product it supports, read the information in “Notices” on
page ix.
This edition applies to Version 7, Release 1, Modification 0 of IBM i (5770-SS1) and related licensed porgram
products.
Notices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix
Trademarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .x
Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiii
The team who wrote this book . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiv
Now you can become a published author, too! . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xvi
Comments welcome. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xvi
Stay connected to IBM Redbooks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xvii
Chapter 7. Storwize V7000 and SAN Volume Controller Copy Services . . . . . . . . . . 113
7.1 Storwize V7000/SAN Volume Controller storage concepts. . . . . . . . . . . . . . . . . . . . . 114
7.1.1 Hardware overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
7.1.2 Storage virtualization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
7.1.3 I/O processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
7.1.4 Copy Services . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
7.2 Metro Mirror. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
7.2.1 Bandwidth thresholds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
7.2.2 Remote copy relationship states . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
7.2.3 Consistency groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
7.3 Global Mirror . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
7.4 FlashCopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
7.4.1 I/O indirection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
7.4.2 Background copy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
7.4.3 FlashCopy relationship states . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
7.4.4 Thin-provisioned FlashCopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
7.4.5 Multi-target and reverse FlashCopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
Contents v
13.1.3 Configuring IBM i DS8000 Global Mirror (CL commands) . . . . . . . . . . . . . . . . 311
13.1.4 Configuring IBM i DS8000 LUN-level switching . . . . . . . . . . . . . . . . . . . . . . . . 317
13.2 Managing IBM i DS8000 Copy Services . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 320
13.2.1 Switchover and switchback for a Metro Mirror or Global Mirror planned outage 320
13.2.2 Using CL commands for DS8000 LUN-level switching . . . . . . . . . . . . . . . . . . . 346
13.2.3 Failing over and back for an unplanned outage . . . . . . . . . . . . . . . . . . . . . . . . 350
13.2.4 Detaching and reattaching a remote copy ASP session . . . . . . . . . . . . . . . . . . 362
13.2.5 Managing FlashCopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 366
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 461
Contents vii
viii PowerHA SystemMirror for IBM i Cookbook
Notices
This information was developed for products and services offered in the U.S.A.
IBM may not offer the products, services, or features discussed in this document in other countries. Consult
your local IBM representative for information on the products and services currently available in your area. Any
reference to an IBM product, program, or service is not intended to state or imply that only that IBM product,
program, or service may be used. Any functionally equivalent product, program, or service that does not
infringe any IBM intellectual property right may be used instead. However, it is the user's responsibility to
evaluate and verify the operation of any non-IBM product, program, or service.
IBM may have patents or pending patent applications covering subject matter described in this document. The
furnishing of this document does not give you any license to these patents. You can send license inquiries, in
writing, to:
IBM Director of Licensing, IBM Corporation, North Castle Drive, Armonk, NY 10504-1785 U.S.A.
The following paragraph does not apply to the United Kingdom or any other country where such
provisions are inconsistent with local law: INTERNATIONAL BUSINESS MACHINES CORPORATION
PROVIDES THIS PUBLICATION "AS IS" WITHOUT WARRANTY OF ANY KIND, EITHER EXPRESS OR
IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF NON-INFRINGEMENT,
MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. Some states do not allow disclaimer of
express or implied warranties in certain transactions, therefore, this statement may not apply to you.
This information could include technical inaccuracies or typographical errors. Changes are periodically made
to the information herein; these changes will be incorporated in new editions of the publication. IBM may make
improvements and/or changes in the product(s) and/or the program(s) described in this publication at any time
without notice.
Any references in this information to non-IBM websites are provided for convenience only and do not in any
manner serve as an endorsement of those websites. The materials at those websites are not part of the
materials for this IBM product and use of those websites is at your own risk.
IBM may use or distribute any of the information you supply in any way it believes appropriate without incurring
any obligation to you.
Information concerning non-IBM products was obtained from the suppliers of those products, their published
announcements or other publicly available sources. IBM has not tested those products and cannot confirm the
accuracy of performance, compatibility or any other claims related to non-IBM products. Questions on the
capabilities of non-IBM products should be addressed to the suppliers of those products.
This information contains examples of data and reports used in daily business operations. To illustrate them
as completely as possible, the examples include the names of individuals, companies, brands, and products.
All of these names are fictitious and any similarity to the names and addresses used by an actual business
enterprise is entirely coincidental.
COPYRIGHT LICENSE:
This information contains sample application programs in source language, which illustrate programming
techniques on various operating platforms. You may copy, modify, and distribute these sample programs in
any form without payment to IBM, for the purposes of developing, using, marketing or distributing application
programs conforming to the application programming interface for the operating platform for which the sample
programs are written. These examples have not been thoroughly tested under all conditions. IBM, therefore,
cannot guarantee or imply reliability, serviceability, or function of these programs.
The following terms are trademarks of the International Business Machines Corporation in the United States,
other countries, or both:
Active Memory™ i5/OS® PowerHA®
AIX® IBM® PowerVM®
AS/400® iCluster® POWER®
DB2® iSeries® Redbooks®
Distributed Relational Database Micro-Partitioning® Redbooks (logo) ®
Architecture™ OS/400® Storwize®
DRDA® POWER Hypervisor™ System i®
DS6000™ Power Systems™ System Storage®
DS8000® POWER6+™ System x®
FlashCopy® POWER6® System z®
Global Technology Services® POWER7® Tivoli®
Microsoft, Windows, and the Windows logo are trademarks of Microsoft Corporation in the United States,
other countries, or both.
Java, and all Java-based trademarks and logos are trademarks or registered trademarks of Oracle and/or its
affiliates.
SAP, and SAP logos are trademarks or registered trademarks of SAP AG in Germany and in several other
countries.
Intel, Intel logo, Intel Inside, Intel Inside logo, Intel Centrino, Intel Centrino logo, Celeron, Intel Xeon, Intel
SpeedStep, Itanium, and Pentium are trademarks or registered trademarks of Intel Corporation or its
subsidiaries in the United States and other countries.
Linux is a trademark of Linus Torvalds in the United States, other countries, or both.
Other company, product, or service names may be trademarks or service marks of others.
Download
Android
iOS
Now
IBM® PowerHA® SystemMirror for i is the IBM high-availability disk-based clustering solution
for the IBM i 7.1 operating system. When combined with IBM i clustering technology,
PowerHA for i delivers a complete high-availability and disaster-recovery solution for your
business applications running in the IBM System i® environment. PowerHA for i enables you
to support high-availability capabilities with either native disk storage or IBM DS8000® or
DS6000™ storage servers or IBM Storwize® V7000 and SAN Volume Controllers.
The latest release of IBM PowerHA SystemMirror for i delivers a brand new web-based
PowerHA graphical user interface that effectively combines the solution-based and
task-based activities for your HA environment, all in a single user interface.
This IBM Redbooks® publication gives a broad understanding of PowerHA for i. This book is
divided into three major parts:
Part 1, “Introduction and background” on page 1, provides a general introduction to
clustering technology, independent ASPs, PowerHA SystemMirror products, and PowerHA
Architecture.
Part 2, “Concepts and planning” on page 65, describes and explains the various interfaces
that PowerHA for i has. It also explains the HA concepts as they pertain to Geographic
Mirroring, DS8000 Copy Services, Storwize V7000 and SAN Volume Controller Copy
Services, and Advanced Copy Services. It also shows you licensing and ordering
information and outlines several considerations to remember when sizing and
performance of the HA solution that you are planning to deploy using IBM PowerHA
SystemMirror for i.
Part 3, “Implementation examples and best practices” on page 197, walks you through
several scenarios with a step-by-step approach for configuring and managing your
IBM PowerHA SystemMirror for i solution. For each scenario, we show you how to
perform a planned switchover, and we also discuss the procedures for unplanned failover.
In Chapter 15, “Best practices” on page 397, we share our recommendations and best
practices to follow in a Highly Available environment that uses various components of the
IBM PowerHA SystemMirror for i product.
If you are new to high availability, we recommend that you follow the general structure and
flow of this book so that you can start by learning the concepts and progress into the
implementation scenarios.
If you are familiar with high availability building blocks or have previous experience with any
of the solutions that we discuss in this book, then we recommend that you familiarize yourself
with the changes that are new to the latest release of the product.
Since the original writing of this book, IBM iCluster® for IBM Power Systems™, a logical
replication software product for high availability and disaster recovery that runs on the IBM i
operating system, has been acquired by Rocket Software, Inc. HA Assist, a derivative offering
of the iCluster product, was also included in the sale to Rocket Software. HA Assist for IBM i
is a specially priced and packaged offering of iCluster, which can be used in conjunction with
PowerHA SystemMirror for i for specific situations.
Hernando Bedoya is a Senior IT Specialist at STG Lab Services and Training, in Rochester,
Minnesota. He writes extensively and teaches IBM classes worldwide in all areas of DB2® for
i. Before joining STG Lab Services he worked in the ITSO for nine years writing multiple IBM
Redbooks publications. He also worked for IBM Colombia as an IBM AS/400® IT Specialist
doing pre-sales support for the Andean countries. He has 25 years of experience in the
computing field and has taught database classes in Colombian universities. His areas of
expertise are database technology, performance, and data warehousing. He has a master’s
degree in Computer Science from EAFIT, Colombia.
Abdel Ali-Darwish is a IT Specialist for IBM i and a Technical Solution Architect working in
the IBM Global Technology Services® support organization in Lima, Perú. He has eight years
of experience with IBM i systems. Over the years he has participated in several pre-sales and
post sales support roles including the midrange platform. He currently provides technical
pre-sales support to IBM Multicountry for Power Systems with emphasis on IBM i. His current
responsibilities include designing solutions and configurations, providing marketing
information, and serving as a subject matter expert in technical and delivery assessments.
Ingo Dimmer is an IBM Consulting IT Specialist for IBM i and a PMI Project Management
Professional working in the IBM STG Europe storage support organization in Mainz,
Germany. He has 12 years of experience in enterprise storage support from working in IBM
post-sales and pre-sales support. His areas of expertise include IBM i external disk and tape
storage solutions, I/O performance, and high availability. He has been an author of several
white paper and IBM Redbooks publications. He holds a degree in electrical engineering from
the Gerhard-Mercator University Duisburg.
Sabine Jordan is a Consulting IT Specialist working in IBM Germany. She has worked as a
Technical Specialist in the IBM i area for more than 20 years, specializing in high availability
since 2004. She has worked on IBM PowerHA System Mirror for i implementations for both
SAP and non-SAP environments using geographic mirroring and DS8000 remote copy
services. Among these implementations, she has created concepts for the design and
implemented the entire project (cluster setup, application changes), in addition to performing
customer education and testing. In addition, Sabine presents and delivers workshops (internal
and external) on IBM PowerHA System Mirror for i and high availability and disaster recovery.
KyoSeok Kim is a Senior IT Specialist for IBM i working for IBM Global Technology Services
in Korea. He has 17 years of experience with AS400, iSeries®, System i, i5/OS®, and IBM i.
He is second-level support (Top Gun) for IBM i and Power System for IBM Korea. His areas of
expertise include internal, performance, database, and IBM i problem determination. He
holds a master’s degree in computer engineering and a Bachelor of Science degree in
physics from Korea University.
Akinori Mogi is an IBM Consulting IT Specialist for IBM i and the Power Systems platform.
He works in the Technical Sales division in IBM Japan. He has over 20 years of experience in
AS/400, iSeries, System i, and IBM i pre-sales and post sales support activities as a technical
expert. His areas of expertise are high-availability solutions, virtualization, and system
performance of the IBM i environment. He has recently joined Power Systems Technical
Sales is responsible for ATS and FTSS. Akinori is also an instructor of information technology
at a university.
Marc Rauzier is an IBM Certified IT Specialist working for IBM France in Global Technology
Services - Services Delivery organization in Lyon. He has more than 20 years of experience
in information technology, focusing on AS/400, iSeries, System i, i5/OS, and IBM i. He is
responsible for the architecture, design, and implementation of IBM Power Systems and the
IBM i based-solution for strategic outsourcing of customers in France. His areas of expertise
include IBM i, HMC, VIOS, SAN, 35xx series tape libraries, and DS8x00 series external
storage related to the IBM i environment.
Figure 1 From left to right: Ingo Dimmer, Akinori Mogi, Tomasz Piela, Sabine Jordan, Abdell
Ali-Darwish, Nandoo Neerukonda, Marc Rauzier, and KyoSeok Kim
Jenifer Servais
Linda Robinson
International Technical Support Organization
Troy Biesterfeld
Tom Crowley
Jenny Dervin
Steven Finnes
Amanda Fogarty
James Lembke
Curt Schemmel
Christopher Wilk
Ben Rabe
IBM Development Rochester
Laural Bauer
Selwyn Dickey
Jerry Evans
Preface xv
Brian Walker
Tim Klubertanz
John Stroh
Cindy Mestad
Steven Ransom
John Stroh
Charles Farrell
IBM STG Lab Services Rochester
Find out more about the residency program, browse the residency index, and apply online at:
ibm.com/redbooks/residencies.html
Comments welcome
Your comments are important to us!
We want our books to be as helpful as possible. Send us your comments about this book or
other IBM Redbooks publications in one of the following ways:
Use the online Contact us review Redbooks form found at:
ibm.com/redbooks
Send your comments in an email to:
[email protected]
Mail your comments to:
IBM Corporation, International Technical Support Organization
Dept. HYTD Mail Station P099
2455 South Road
Poughkeepsie, NY 12601-5400
Preface xvii
xviii PowerHA SystemMirror for IBM i Cookbook
Part 1
IBM recently made significant investments to further enhance PowerHA SystemMirror for i as
its strategic high availability product for the IBM i platform.
With the October 2011 announcement, PowerHA SystemMirror for i now also supports
IBM System Storage® SAN Volume Controller and IBM Storwize V7000 for storage-based
replication solutions. A new PowerHA GUI and further enhancements for IASP-based
high-availability solutions complement the new PowerHA functionality.
This book is structured in three parts, with Part 1 covering an introduction and the architecture
of IBM PowerHA SystemMirror for i, Part 2, “Concepts and planning” on page 65, providing
concepts and information about planning, and Part 3, “Implementation examples and best
practices” on page 197, providing implementation examples and scenarios along with
best practices.
An unplanned outage that in duration or recovery time exceeds business expectations can
have severe implications, including unexpected loss of reputation, customer loyalty, and
revenue. Customers who did not effectively plan for the risk of an unplanned outage, never
completed their installation of an HA solution, or did not have a tested tape recovery plan in
place are especially exposed to negative business impacts.
Figure 1-1 shows an example how a high-availability solution can help you to significantly
reduce planned and unplanned application downtimes.
unplanned
outage Failure Reload Restore Journal Partial Reconcile
(without HA) Started continues Recovery Production Journal Data
unplanned
outage Move Users < Active Production Resync HA to Prod
Repair
with HA to HA Server on HA Server > Move users back to Prod
planned
Move Users Power Off Production < Active Production Resync HA to Prod
outage
to HA Server Server for Maintenance on HA Server > Move users back to Prod
with HA
Outage
Outage comparisonwith
comparison withand
and without
without HA
HA
Unavailable Uptime
Switched IASPs Geographic Mirroring Metro Mirror Global Mirror FlashCopy LUN Level Switching1
• Single copy of • IBM i storage mgmt • SAN hardware • SAN • SAN • Single copy of IASP
IASP is switched page level replication replication hardware hardware switched between
between LPARs • Sync. or asynchronous1 • Synchronous replication replication LPARs or server
• Internal or • Internal or external • IBM DS8000/ • Asynchronous • Point in time • IBM DS8000/
external storage storage DS6000 or • IBM DS8000/ • IBM DS8000/ DS6000
• Supports direct, VIOS & SVC/V7000 DS6000 or DS6000 or
IBM i hosted storage SVC/V7000 SVC/V7000
1
requires IBM i 7.1 or later
SVC/V7000 storage-based replication requires IBM i 7.1 with 5799-HAS PRPQ
The key characteristic that our PowerHA customers love is that the solution is automated.
Because the data resiliency is completely managed within the IBM i storage management
architecture, there is no operator involvement, just as there is no operator involvement with
RAID 5 or disk mirroring. Geographic mirroring offers IBM i customers an IBM i-based
page-level replication solution for implementing high availability and disaster recovery with
any kind of IBM i-supported internal or external storage solution. With IBM System Storage
DS8000/DS6000 or SAN Volume Controller (SVC)/Storwize V7000 storage servers our
clients are able to exploit storage-based remote replication functions for high availability and
disaster recovery, LUN-level switching for local high availability, and FlashCopy® for reducing
save window outages by enabling the creation of a copy that is attached to a separate
partition for off-line backup to tape.
DS8000 DS8000
LPAR-1
Production HA
LPAR-1 LPAR-1
*SYSBAS Metro
*SYSBAS IASP Mirror IASP *SYSBAS
IASP *SYSBAS *SYSBAS
*SYSBAS
Global Mirror LUN-level Switching PowerHA SystemMirror for I combined solution with
System Storage Copy Services using
FlashCopy, LUN level switching and remote replication
DS8000
For more detailed information about IBM PowerHA SystemMirror for i architecture and
business resiliency solutions see Chapter 4, “PowerHA architecture” on page 45.
PowerHA BC Tier 7 – Server or Storage replication with end-to-end automated server recovery
Metro /
Cost / Value
When you start thinking about implementing HA for your IBM i environment, consider how this
criteria apply to your situation before deciding which solution fits your needs best:
Types of outages to be addressed
– Unplanned outages (for example, a hardware failure)
– Planned outages (for example, a software upgrade)
– Backups (for example, creating a copy of disk for an online save to tape)
– Disasters (for example, site loss, power grid outage, and so on)
Recovery objectives
– Recovery time objective (RTO): The time to recovery from an outage
– Recovery point objective (RPO): The amount of tolerable data loss (expressed as a
time duration)
iCluster
Asynchronous
APPLICATION APPLICATION Journal-based replication
Logical
over TCP/IP
replication IBM i IBM i Disk subsystem agnostic
Synchronous or Asynchronous
Geographic
OS-based APPLICATION APPLICATION IBM i-based replication
Mirroring
over TCP/IP
replication IBM i IBM i Disk subsystem agnostic
Synchronous or Asynchronous
Storage-based replication
Storage-based APPLICATION APPLICATION controlled by DS8000
over Fibre Channel or FCIP for
replication IBM i IBM i DS8000 and
Metro Mirror SVC/V7000
Global Mirror
Solution considerations
In this section we explain concepts that can help you to decide which solution your
business requires.
The cluster administrative domain is the PowerHA function that ensures that the set of
objects that are not in an IASP are synchronized across the nodes in the cluster. Thus, the
application has the resources that it needs to function on each node in the cluster. Clustering
solutions deployed with iASPs and using either storage-based copy services or geographic
mirroring replication require little in the way of day-to-day administrative maintenance and
were designed from the beginning for role-swap operations. We define an HA environment as
one in which the primary and secondary nodes of the cluster switch roles on a regular and
sustained basis.
Rule of thumb: If your business does not conduct regular and sustained role swaps, your
business does not have a high-availability solution deployment.
The logical replication in the IBM i environment is based on IBM i journaling technology,
including the option of remote journaling. A key characteristic of logical replication is that only
those objects that are journalled by IBM i (that is, database, IFS, data area, data queue) can
be replicated in near real time.
In addition, because one can choose to replicate a subset of objects, the bandwidth
requirements are typically not as great in comparison to a hardware-based replication approach.
As the need to segregate groups of programs and data on the same system emerged, the
concept of pools developed and was included as part of the operating system. The pools
were referred to as auxiliary storage pools (ASPs) because they pertained to areas of
auxiliary storage (disk space). The new command structures within the operating system
used the letters ASP when referring to the auxiliary storage pools.
Enhancements to the concept of pools has led to independent auxiliary storage pools
introduced with OS/400® V5R1. These are pools that can be brought online, taken offline,
and accessed independently of the other pools on the system. They can even be logically or
physically switched between systems or logical partitions.
ASPs
System User
SYSBAS
Basic Independent
2-32 33-255
UDFS
Primary
Disk pool
group Secondary
Secondary
There is a difference between basic ASPs and independent ASPs when it comes to data
overflow. Basic ASPs overflow and independent ASPs do not. An overflow of a basic user
ASP occurs when the ASP fills. The excess data spills into the system ASP. IASPs are
designed so that they cannot overflow. Otherwise, they would not be considered independent
or switchable. An IASP is allowed to fill up, and the application that is responsible for filling it
up simply halts. There is no automatic cancellation of the responsible job. If this job is running
from a single-threaded JOBQ, in a single-threaded subsystem all further processing is
stopped until user action is initiated.
When an IASP fills up, the job that generates the data that filled up the disk pool might not be
complete. The system generates an MCH2814 message indicating this condition. This might
have serious ramifications. Jobs that only read data are still able to work, but any job trying to
add data to the IASP is on hold. The system does not automatically cancel the offending job.
If the job is from a single-threaded JOBQ or a single-threaded subsystem, other jobs behind it
are held up until the offending job is handled. Possible scheduling impacts might occur.
With library-capable IASPs, a thread can reference, by name, all of the libraries in the IASPs
of one ASP group. This adds a second, but optional, component to the name space and is
referred to as the ASP group component of the name space. A thread that does not have an
ASP group component in its name space has its library references limited to the *SYSBAS
component. A thread with an ASP group component to its library name space can reference
libraries in both the *SYSBAS and the ASP group components of its name space.
Library names no longer must be unique on a system. However, to avoid ambiguity in name
references, library names must be unique within every possible name space. Because
*SYSBAS is a component of every name space, presence of a library name in *SYSBAS
precludes its use within any IASP. Because all libraries in all IASPs of an ASP group are part
of a name space, for which the ASP group is a component, existence of a library name within
one IASP of an ASP group precludes its use within any other IASP of the same ASP group.
Because a name space can have only one ASP group component, a library name that is not
used in *SYSBAS can be used in any or all ASP groups.
IBM i has a file interface and an SQL interface to its databases. The file interface uses the
name space to locate database objects. For compatibility, SQL maintains a catalog for each
ASP group. This catalog resides in the primary IASP of the ASP group. The catalog is built
from the objects that are in a name space that has the ASP group and *SYSBAS as its two
components. The names database and the name space are somewhat interchangeable
because they refer to the same set of database objects.
Each name space is treated as a separate relational database by SQL. It is required that all
RDBs whose data is accessible by SQL are defined in the RDB directory on the system.
Note that the name space is a thread attribute and can be specified when a job is started.
When it is referenced as a job attribute, it technically means the “thread attribute for the initial
thread of a single-threaded job.”
Each IBM i system in the distributed relational database network must have a relational
database directory configured. There is only one relational database directory on a system.
Each AR in the distributed relational database network must have an entry in its relational
database directory for its local RDB and one for each remote and local user RDB that the AR
accesses. Any system in the distributed RDB network that acts only as an application server
does not need to include the RDB names of other remote RDBs in its directory.
Figure 2-2 gives an example of the RDB directory on a system with an IASP configured.
Notice that there is one entry present with a remote location of local. This is the RDB entry
representing the database in SYSBASE. In addition, an RDB entry gets created by the
operating system when you vary on an IASP. In our case this is the entry IASP1 with a
remote location of loopback. By default, the relational database name of an IASP is identical
to the IASP device name, but you can also choose another name here. When migrating an
application environment with a large number of accesses through the RDB name, you might
want to change the SYSBASE RBD name to a different value and use the “old” SYSBASE
RDB name as the database name for the IASP. This way, you do not have to change RDB
access to your environment.
Position to . . . . . .
Remote
Option Entry Location Text
Bottom
F3=Exit F5=Refresh F6=Print list F12=Cancel F22=Display entire field
(C) COPYRIGHT IBM CORP. 1980, 2009.
Figure 2-2 Work with Relational Database Directory Entries
Although the objects in the system RDB are logically included in a user RDB, certain
dependencies between database objects have to exist within the same RDB. These include:
A view into a schema must exist in the same RDB as its referenced tables, views,
or functions.
An index into a schema must exist in the same RDB as its referenced table.
A trigger or constraint into a schema must exist in the same RDB as its base table.
Parent table and dependent table in a referential constraint both have to exist in the
same RDB.
A table into a schema has to exist in the same RDB as any referenced distinct types.
2.1.3 Connections
In an SQL environment, SQL CONNECT is used to specify the correct database. To achieve
the best performance, make sure that the database being connected to corresponds with
your current library name space. You can use SETASPGRP or the INLASPGRP parameter in
your job description to achieve this. If the SQL CONNECT function is not operating within the
same library name space, the application uses Distributed Relational Database
Architecture™ (DRDA®) support, which can affect performance.
Consider the example in Example 2-1 with library demo10 residing in an IASP and the job
running the SQL being attached to an IASP. In this example, the view ICTABLES is not built in
the current library (DEMO10) as you would expect. It is built in the library of the first table that is
mentioned, which is QSYS2 (where SYSTABLES is located). It fails when accessing the
independent disk pool because creation of objects in QSYS2XXXXX is prevented. In the
example mentioned, you must explicitly specify that you want to create the view either in
QSYS2 or in a user library in the IASP.
The IASP version of QSYS and QSYS2 contains cross-reference and SQL catalog tables
with merged views of all the SQL and database objects that are accessible when connected
to the IASP.
You are presented with the current disk pool configuration (Figure 2-5).
3. In the next window, choose Primary as the type of disk pool (Figure 2-7). You can then
enter the name of your IASP. The database name defaults to the name of the IASP, but
you can also enter a different name for the database. Be aware that the IASP database
name cannot be identical to the system ASP database name.
5. Choosing Add Disks to be Mirrored then gives us a list of disks that can be put into the
new IASP (Figure 2-9). Make sure to select all the disks that you want to have in your
IASP and click Add.
7. The IASP creation then starts (Figure 2-11). Be aware that this screen does not
automatically refresh. Refresh must be done manually.
In 11.3, “Creating an IASP” on page 215, we show how to create an IASP using the new
PowerHA GUI.
Bottom
F1=Help F9=Calculate Selection F11=View 2 F12=Cancel
Specifying *SELECT for the disk unit parameter shows the screen shown in Figure 2-13. It
provides you with a list of unconfigured disks on your system. Choose which ones you want to
add to your IASP and press Enter to create the IASP.
Resource
Opt Name Serial Number Type Model Capacity Rank Eligible
1 DD007 YQKJGD54BUK6 6B22 0050 19088 002 Yes
1 DD006 YDP4V2FVUK63 6B22 0050 19088 002 Yes
1 DD008 YUNHA7W9URJL 6B22 0050 19088 002 Yes
1 DD005 YWPZGH6N8LA9 6B22 0050 19088 002 Yes
Bottom
F1=Help F9=Calculate Selection F11=View 2 F12=Cancel
Configuration of ASP device IASP1 is 8% complete.
Figure 2-13 CFGDEVASP command: Select disks to put into IASP
A message on the bottom of the screen shows the progress of the IASP creation. Your
screen is locked as long as the creation of the IASP is running. You can also follow the status
of the IASP creation using DSPASPSTS.
For some of these object types, special considerations have to be taken into account:
If network attributes reference the alert table, then this object must reside in the
system ASP.
If an active subsystem references a class object, then this class object must reside in the
system ASP.
Database files that are either multiple-system database files or that have DataLink fields
that are created as link control cannot be located in an independent disk pool. If an active
subsystem references the file object, *FILE must exist in the system disk pool, for
example, the sign-on display file.
Subsystem descriptions can be placed into in IASP. However, if you want to start a
subsystem, its subsystem description has to be in the system ASP at that point in time.
Table 2-2 shows which object types cannot be put into an IASP.
If you look at the list of object types you will notice that most of them are either legacy
objects (such as folders or documents), configuration objects (such as device descriptions or
line descriptions), or objects closely related to system security (such as user profiles or
authority lists).
If you want to keep objects in the system ASP in sync between your production and your
backup system (such as user IDs and passwords) you can use the administrative domain to
help you with this task. See 3.5, “Advanced node failure detection” on page 43.
The recommended way is to use job descriptions, as that is also applicable to prestarted jobs.
Important: Never change the QDFTJOBD job description. Doing so leaves you with an
unusable system after an IPL.
For more information about these commands, see the CL Command Finder function in the
iSeries Information Center:
http://publib.boulder.ibm.com/eserver/ibmi.html
If for any reason using a specific job description is not possible with some ODBC or JDBC
connections, then they both also provide parameters in the connection setup to explicitly
include an IASP in their namespace (Example 2-2) for ODBC.
If this does not work in some cases in your environment (for example, because you provide
anonymous FTP access to your system), then you can use the following command to get
access to IASP1 within the FTP job:
quote rcmd SETASPGRP IASP1
Access to IFS objects is not affected by SETASPGRP or by a job description pointing to an IASP
but has to be done using the hierarchical path structure of the IFS. Therefore, if you do not
want to change hard-coded paths to IFS objects in your applications, you can create symbolic
links from the original location to the IASP location.
Tip: Check that you do not have any IFS directories with the same name as the primary
IASP that you want to create.
Be aware that the process of varying on an IASP causes the RDB directory to be unavailable
for a short period of time. This can cause attempts by a DRDA application requester or
application server to use the directory to be delayed or to time out.
Local user database entries in the RDB directory are added automatically the first time that
the associated databases are varied on. They are created using the *IP protocol type and
with the remote location designated as LOOPBACK. LOOPBACK indicates that the database
is on the same server as the directory.
When a static SQL program is run, an access plan is created and stored with the program. If
the SQL program is located in the system ASP, a job or thread with IASP1 in its namespace
will create an access plan for data in IASP1. When a job or thread with IASP2 in its
namespace runs the same SQL program, the existing access plan is invalidated, so a new
access plan is created and stored in the program. We therefore recommend creating
separate static SQL applications in each IASP for best performance if you use more than one
IASP in your environment.
For extended dynamic SQL, create a separate SQL package in each IASP for best performance.
When output from STRQMQRY is directed to an output file, Query Management ensures that
the output file is created on the RDB (name space) that was current at the time that STRQMQRY
is executed.
CLUSTER
Node 1 Node 2
Note: Small outages, tolerated just a few years ago, can now mean a significant loss of
revenue and of future opportunities for a business. The most important aspect of clustering
is high availability (that is, the ability to provide businesses with resilient resources).
Cluster
Collection of IBM i Systems
Device Domain
Collection of cluster nodes that share resources (switchable DASD towers)
Manages assignment of common IASP ID, disk unit and virtual addresses across domain
Device Description
Logical control name for varying on/off
an IASP
Drives Drives
IASP
Defines a physical set of switchable
drives
Pre-requisite: cluster
Clusters are a very effective solution for continuous availability requirements on an IBM i
system or logical partition, providing fast recovery for the widest range of outages possible,
with minimal cost and overhead.
Some people might think that a backup of the server is not an outage. But IBM i users are not
interested in such technicalities. If access to their data on the system is not possible, the user
is most concerned about when the system is available again so that work can continue.
IBM i clustering technology offers you state-of-the-art and easy-to-deploy mechanisms to put
your business on the path to continuous availability (Figure 3-3).
Administrative
Domian
IBM i Cluster
Scalability RAS - Reliability, Availability, Serviceability
Automation
Backup and Recovery Concurrent Maintenance
PowerHA for i
Selective Fault Tolerance end-to-end
Systems Management solutions
Fault Isolation
Performance Management Predictive Failure Analysis
Any name can be given to a node. However, we recommend that you make the node
name the same as the system name. Cluster communications that run over IP connections
provide the communications path between cluster services on each node in the cluster. The
set of cluster nodes that is configured as part of the cluster is referred to as the cluster
membership list.
A cluster consists of a minimum of two nodes. The environment can be extended to a cluster
with a maximum of 128 nodes.
Device Domain
Node 1 Node 2
Cluster Resources
Device Domain
These cluster resources are negotiated across a device domain to ensure that there are
no conflicts:
IASP number assignments
IASPs are automatically assigned a number to correlate the name of the IASP. The user
chooses the resource name. The system manages the assigned IASP numbers, which
might not be in numerical order. The order depends on a number of factors, including the
creation date and the creation of IASPs on other nodes in the device domain.
DASD unit number assignments
To keep from conflicting with the permanently attached disk units of each node, all IASP
unit numbers begin with a four.
Virtual address assignments
The cluster configuration determines the virtual address space required for the IASP.
Virtual address assignments (the cluster configuration) are ensured not to conflict across
all nodes in the device domain.
Note: The collection of switched disks, the independent disk pool identification, disk
unit assignments, and virtual address assignments must be unique across the entire
device domain.
Resources that are available or known across multiple nodes within the cluster are called
cluster resources. A cluster resource can conceptually be any physical or logical entity (that
is, database, file, application, device). Examples of cluster resources include IBM i objects, IP
addresses, applications, and physical resources. When a cluster resource persists across an
outage, that is any single point of failure within the cluster, it is known to be a resilient
resource. As such, the resource is resilient to outages and accessible within the cluster even
if an outage occurs to the node currently hosting the resource.
Cluster nodes that are grouped together to provide availability for one or more cluster
resources are called the recovery domain for that group of cluster resources. A recovery
domain can be a subset of the nodes in a cluster, and each cluster node might participate in
multiple recovery domains. Resources that are grouped together for the purposes of recovery
action or accessibility across a recovery domain are known as a cluster resource group
(Figure 3-5).
CRG
B
Cluster Resources
(for example, switched disk with IASP)
There are four cluster resource group (CRG) object types that are used with Cluster Services
at V7R1:
Application CRG
An application CRG enables an application (program) to be restarted on either the same
node or a different node in the cluster.
Data CRG
A data CRG enables data resiliency so that multiple copies of data can be maintained on
more than one node in a cluster.
The cluster resource group defines the recovery or accessibility characteristics and behavior
for that group of resources. A CRG describes a recovery domain and supplies the name of
the cluster resource group exit program that manages cluster-related events for that group.
One such event is moving the users from one node to another node in case of a failure.
Recovery domain
A recovery domain is a subset of nodes in the cluster that are grouped together in a cluster
resource group for purposes such as performing a recovery action. Each cluster resource
group has a recovery domain that is a subset of the nodes in the cluster. Here are facts about
recovery domains:
The nodes within a recovery domain participate in any recovery actions for the resources
of the domain.
Different CRGs might have different recovery domains.
As a cluster goes through operational changes (for example, nodes end, nodes start,
nodes fail), the current role of a node might change. Each node has a preferred role that is
set when the CRG is created.
A recovery domain can be a subset of the nodes in a cluster, and each cluster node might
participate in multiple recovery domains.
An exit program is called when a CRG detects certain events, such as a new node being
added to the recovery domain, or the current primary node failing. The exit program is called
with an action code indicates what the event is. Furthermore, the exit program has the
capability to indicate whether to process the event. User-defined simply means that the IBM i
cluster technology does not provide the exit program. Typically the exit program is provided
by the application or data replication provider. The exit program is the way that a CRG
communicates cluster events to the exit program provider. The exit program can perform the
appropriate action based on the event, such as allowing a resource access point to move to
another node. The exit program is optional for a resilient device CRG but is required for the
other CRG types. When a cluster resource group exit program is used, it is called on the
occurrence of cluster-wide events.
For detailed information about the cluster resource group exit programs, including what
information is passed to them for each action code, see:
http://publib.boulder.ibm.com/infocenter/iseries/v7r1m0/topic/apis/clrgexit.htm
With IBM i 7.1, PowerHA SystemMirror for i now allows advanced node failure detection by
cluster nodes. This is done by registering with an HMC or Virtual I/O Server (VIOS)
management partition on IVM-managed systems. This way clustering gets notified in case of
a severe partition or system failure to trigger a cluster failover event instead of causing a
cluster partition condition.
If a node detects a failure, it notifies the other nodes in the cluster via a distress message.
This triggers a failover if the failing node is the primary node of a CRG. If instead we get a
heartbeating failure, then a partition occurs. Partitions are usually the result of network
failures, and the nodes automatically merge after the problem has been fixed. However, there
are cases in which the node fails too quickly for a distress message to go out, which can
result in a “false” partition. The user can use CHGCLUNODE to change a partitioned node to a
failed node. Through the use of advanced node failure detection, the occurrences of false
partitions are reduced.
For LPAR failure conditions it is the POWER® Hypervisor™ (PHYP) that notifies the HMC
that a LPAR failed. For system failure conditions other than a sudden system power loss, it is
the flexible service processor (FSP) that notifies the HMC of the failure. The CIM server on
the HMC or VIOS can then generate an IBM Power state change CIM event for any
registered CIM clients.
Important: For systems that can be managed with a Hardware Management Console
(HMC), the Common Information Model (CIM) server runs on the HMC. The HMC affords
the most complete node failure detection because it is not part of the system and thus can
continue to operate when a system has completely failed. When using VIOS, the CIM
server must be started in the VIOS management partition by running startnetsvc
cimserver in the VIOS partition.
Whenever a cluster node is started, for each configured cluster monitor IBM i CIM client APIs
are used to subscribe for the particular power state change CIM event. The HMC CIM server
generates such a CIM event and actively sends it to any registered CIM clients (that is, there
is no heartbeat polling involved with CIM). On the IBM i cluster nodes the CIM event listener
compares the events with available information about the nodes constituting the cluster to
determine if it is relevant for the cluster to act upon. For relevant power state change CIM
events, the cluster heartbeat timer expiration is ignored (that is, IBM i clustering immediately
triggers a failover condition in this case).
Clustering solutions build on PowerHA, based on the technologies of the physical exchange
of the data included on the independent ASP (iASP). These are the mechanisms used by
PowerHA described here:
Switched disks
Host-based replication
Storage-based replication
Administrative domain
Figure 4-1 and Figure 4-2 on page 47 provide information about PowerHA SystemMirror
feature availability in relation to the hardware being used.
1 Requires NPIV capable fiber channel adapter. DS5000 NPIV support requires IBM i 7.1 TR2
2 Requires IBM i 7.1 TR3 and PRPQ 5799-HAS
3 See Redbook “IBM i and Midrange External Storage SG247668” and ”IBM i Virtualization and DS4000 Read-me First”
SYSBAS SYSBAS
Switched
Disk
Node1 Node2
IASP
Site A
Figure 4-3 Model of switched disk solution
Note: POWER6® server is the last one that uses the HSL loop and the last server where
switching disks build on this technology is supported.
Using the switched disk solution we have a single IASP that can be owned by one or the
other node in the cluster at a given time. During the switch process the switchable IASP
changes its owner and all the data on it is accessible by the other node.
Switched disks are not advised to be a single HA solution for a production system. However
they can be a complementary element of more sophisticated solutions (for example, a 3-node
solution with storage-based replication).
Note: The data on the switched disks is in one copy, so it should be protected on the disk
level (for example, by one of the RAID mechanisms).
In the case of the PowerHA System Mirror for i, the host-based replication feature is called
geographic mirroring.
PRODUCTION MIRROR
copy copy
SYSBAS SYSBAS
/ Network
TCP/IP
IASP IASP
source Geographic target
Mirroring
Site A Site B
Figure 4-4 Host-based replication in PowerHA: Geographic mirroring
Note: Notice that iASPs can be built on internal or external disk drives.
Geographic mirroring guarantees that the changes to the replicated iASP will be applied to
the target copy in the original order, because in case of failure the target copy will be usable
and accessible by the node that owns it.
Note: It must be stressed that the commitment control must be in place to guarantee the
database consistency in case of unexpected failure.
Geographic mirroring introduces high-availability protection in case of the server failure, and
in case the servers are in a different location, it allows protection in case of a site outage.
The copy owned by the primary node is the production copy and the copy owned by the
backup system at the other site is the mirror copy. Users and applications can only access
the independent disk pool on the primary node, the node that owns the production copy.
Changes that are made to the production copy (source system) are guaranteed by the
geographic mirroring functionality to be made in the same order on the mirror copy
(target system).
Node 1
PRODUCTION
Copy
Node 2
2 MIRROR
Main TCP/IP Network Copy
Memory 3
4 1
Main
Memory
IOA cache
IASP
source
5
IOA cache
IASP
target
Site A Site B
Figure 4-5 Geographic mirroring operations
More details about the geographic mirroring can be found in Chapter 5, “Geographic
Mirroring” on page 67.
Note: When using host-based replication in synchronous mode of operation, the network
latency caused by the distance between the hosts can cause performance degradation of
the applications using the replicated iASP. The mirroring mode can severely affect
performance when the target system’s disk subsystem has lower performance than the
source system.
Disadvantages
The disadvantage of this solution can be the fact that it uses the host resources, such as CPU
and memory. That might affect performance of other processes running on the system.
When using a Metro Mirror replication with PowerHA System Mirror, each node has a locally
attached external IBM Storage System, and the replicated IASP must be located in it.
Figure 4-6 shows the most common configuration. The System ASP can reside on internal or
external disks.
SITE 1
Production Backup
LPAR LPAR
IASP IASP
SRC Metro Mirror CPY
In addition to IBM TotalStorage devices you can build a Metro Mirror solution using IBM
Storwize V7000 and SAN Volume Controller. For more details about this type of
implementation see 7.2, “Metro Mirror” on page 119.
The use of FlashCopy and the FlashCopy consistency group with Global Copy allows you to
maintain a consistent copy of the IASP in the backup site. Figure 4-7 shows the general
architecture for the Global Mirror replication.
SITE 1 SITE 2
Production Backup
LPAR LPAR
IASP IASP
SRC Global Copy CPY1
FlashCopy
IASP
CPY2
How it works
Global Mirror, as a long-distance remote copy solution, is based on an efficient combination
of Global Copy and FlashCopy functions. It is the Storage system microcode that provides,
from the user perspective, a transparent and autonomic mechanism to intelligently utilize
Global Copy in conjunction with certain FlashCopy operations to attain consistent data at the
remote site.
Global Mirror can be used in solutions based on IBM Storwize V7000 and SAN Volume
Controller. See 7.3, “Global Mirror” on page 123.
LUN-level switching
LUN-level switching is a new function provided by PowerHA SystemMirror for i. It allows you
to switch a set of LUNs (a volume group) between systems. The idea of this solution is similar
to the switched disks described in 4.1.1, “Switched disks” on page 47. For more details about
this solution, see 6.4, “LUN-level switching” on page 105.
FlashCopy
FlashCopy allows you to create a point-in-time copy of the logical volume. By doing a
FlashCopy, a relationship is established between a source volume and a target volume. Both
are considered to form a FlashCopy pair. As a result of the FlashCopy, either all physical
blocks from the source volume are copied (when using the copy option) to the target volume,
or when using the nocopy option. Only those parts are copied that are changing in the source
data since the FlashCopy was established. The target volume needs to be the same size or
bigger than the source volume whenever FlashCopy is used to flash an entire volume.
Figure 4-8 shows an example of the architecture for the FlashCopy. In this example, when the
production IASP FlashCopy is established, the copy can be accessed by another LPAR. The
copy of the IASP is read and write capable, so you can use it for backup or testing purposes.
Production
LPAR IASP
SRC
FlashCopy
Backup
LPAR IASP
CPY
Within PowerHA for i, the classic FlashCopy and FlashCopy SE are supported. When using
classic FlashCopy, the target volume size that is reported to the system must be allocated in
the TotalStorage device.
When using FlashCopy SE (Space Efficient) you can create virtual volumes that are space
efficient and use them as a target for the FlashCopy operation. The space efficient volumes
are volumes that have a defined virtual size, but the physical storage is not physically
allocated to them.
Typically, large applications such as databases have their data spread across several
volumes, and their volumes should all be FlashCopied at exactly the same point-in-time.
FlashCopy offers consistency groups, which allows multiple volumes to be FlashCopied at
exactly the same instance.
There are often configuration parameters or data associated with applications and application
data, which are known collectively as the operational environment for the application.
Examples of this type of data include user profiles used for accessing the application or its
data, or system environment variables that control the behavior of the application. With a
high-availability environment, the operational environment needs to be the same on every
system where the application can run, or where the application data resides. When a change
is made to one or more configuration parameters or data on one system, the same change
needs to be made on all systems. A cluster administrative domain lets you identify resources
that need to be maintained consistently across the systems in an IBM i high availability
environment. The cluster administrative domain then monitors for changes to these resources
and synchronizes any changes across the active domain.
When a cluster administrative domain is created, the system creates a peer CRG with the
same name. The nodes that make up the cluster administrative domain are defined by the
CRGs recovery domain. Node membership of the cluster administrative domain can be
modified by adding and removing nodes from the recovery domain using these commands:
Add Admin Domain Node Entry (ADDCADNODE).
Remove Admin Domain Node Entry (RMVCADNODE).
Work with Cluster (WRKCLU).
Each cluster node can be defined in only one cluster administrative domain within the cluster.
Note: To work with cluster CL commands or the Cluster Resource Services graphical
interface, you must have IBM PowerHA SystemMirror for i licensed program installed.
After the cluster administrative domain is created, it can be managed with CL commands or
the Cluster Resource Services graphical interface in IBM Systems Director Navigator for i.
The type of objects that can be managed in a cluster administrative domain, also known as
monitored resources, have been enhanced in IBM i 7.1. See Table 4-1 on page 51:
Classes *CLS
Monitored resources
A monitored resource is a system resource that is managed by a cluster administrative
domain. Changes made to a monitored resource are synchronized across nodes in the
cluster administrative domain and applied to the resource on each active node. Monitored
resources can be system objects such as user profiles or job descriptions. A monitored
resource can also be a system resource not represented by a system object, such as a single
system value or a system environment variable. These monitored resources are represented
in the cluster administrative domain as monitored resource entries (MREs).
A cluster administrative domain supports monitored resources with simple attributes and
compound attributes. A compound attribute differs from a simple attribute in that it contains
zero or more values, while a simple attribute contains a single value. Subsystem Descriptions
(*SBSD) and Network Server Descriptions (*NWSD) are examples of monitored resources
that contain compound attributes.
For MREs to be added, the resource must exist on the node from which the MREs are added.
If the resource does not exist on every node in the administrative domain, the monitored
resource is created. If a node is later added to the cluster administrative domain, the
monitored resource is created. MREs can only be added to the cluster administrative domain
if all nodes in the domain are active and participating in the group. MREs cannot be added in
the cluster administrative domain if the domain has a status of Start of changePartitionedEnd
of change.
You can add the MRE using ADDCADMRE or with the PowerHA GUI. The PowerHA GUI allows
you to select the MRE’s monitored parameters from a list, while in the command you need to
specify them.
To remove MRE from the administrative domain, you can use RMVCADMRE or the
PowerHA GUI.
Note: To use the Cluster Resource Services graphical interface or the Display CRG
Information (DSPCRGINF) command, you must have the IBM PowerHA SystemMirror for i
licensed program installed.
PowerHA
Cluster
Node 1 Node 2
ASP
Session
Figure 4-9 General view of the ASP copy descriptions and ASP sessions
ASPDEV This is the name of the IASP for which the ASP
copy description is being created. One IASP can
have many ASP copy descriptions.
LOCATION This specifies the node that is used for this ASP
copy description.
Copy Descriptions
ASP
device Name Role State Node
IASP1 IASP1CPYD1 SOURCE UNKNOWN NODE1
IASP1 IASP1CPYD2 TARGET ACTIVE NODE2
Bottom
Press Enter to continue
4.3.1 Start/end
To start the session you can issue straspssn from the 5250 session (Figure 4-12 on
page 61). Table 4-4 describes the parameters for this command.
Bottom
F3=Exit F4=Prompt F5=Refresh F12=Cancel F13=How to use this display
F24=More keys
The ASP session is also started as a result of making IASP highly available in the PowerHA
cluster. To stop the session use this command:
ENDASPSSN SSN(IASP1SSN)
Or use the wrkaspcpyd option 24 next to the related ASP copy description. When the ASP
session is ended, the relation between the ASP copy descriptions is removed.
The Change Auxiliary Storage Pool Session (CHGASPSSN) command can be used to change an
existing geographically mirrored, Metro Mirrored, Global Mirrored, or FlashCopy session.
To suspend or resume geographic mirroring, Metro Mirror, or Global Mirror through this
command, you must have already configured the mirror copy disks and cluster resource
group with a recovery domain that has two sites.
To change session attributes (the *CHGATTR parameter on the CHGASPSSN command) when
using geographic mirroring, the production copy of the iASP must be varied off.
CHGASPSSN with the *Detach, *Reattach, *Suspend, and *Resume options can be
confusing. It is hard to know which node you have to run which option from, in addition to
what states the iASP copies must be in first. Table 4-5 provides a quick reference list. To
simplify the chart we are letting source also mean production copy and target also mean
mirror copy.
Table 4-5 What options can be run from where and what status the iASP copies must be in
Source IASP Target IASP
CHGASPSSN Can run from Can run from must be varied must be varied
option Environment source? Target? off off
CHGASPSSN *SUSPEND or *DETACH will offer the tracking *YES or *NO (Figure 4-14).
However, the parameter is ignored if this is not a geographic mirroring solution. Both Metro
Mirror and Global Mirror will track regardless of this option.
Bottom
F3=Exit F4=Prompt F5=Refresh F12=Cancel F13=How to use this display
F24=More keys
Figure 4-14 CHGASPSSN *DETACH for geographic mirroring
CHGASPSSN with geographic mirror will allow a *DETACH if the iASP is varied on. Because
the iASP is in use, this means that the mirror copy will have to go through an abnormal
vary-on, and there are risks to your data.
CHGASPSSN with Metro Mirroring will not allow a *DETACH if the iASP is varied on. If you
try you will get a CPD26B9 (Figure 4-15).
CHGASPSSN using the *DETACH parameter removes a FlashCopy relation, but it keeps the
ASP session.
When you are going to reattach a Metro Mirror session, a message (CPF9898) will go to the
QSYSOPR message queue that you will have to use to confirm the reattach.
The basis of geographic mirroring is extended IBM i disk mirroring technology to multiple
systems environment. Geographic mirroring is managed by IBM i storage management
capability so that replication is performed by each memory page segment (4096 bytes).
Geographic mirroring is intended for use by clustered system environments and uses data
port services. Data port services are System Licensed Internal Code that supports the
transfer of large volumes of data between a source system and one of any specified target
systems. This is a transport mechanism that communicates over TCP/IP.
Prior to IBM i 7.1, it provides only synchronous delivery mode. Be aware that a local write
waits for the data to reach the main storage of the backup node before the write operation is
considered to be finished.
Cluster
Device domain
Cluster resource group (CRG)
Node A Node B
Production Copy Mirror Copy
TCP/IP Network
SYSBAS SYSBAS
IASP IASP
source target
The copy owned by the primary node is the production copy, and the copy owned by the
backup system at the other site is the mirror copy. Users and applications can only access the
independent disk pool on the primary node, the node that owns the production copy. Changes
that are made to the production copy (source system) are guaranteed by the geographic
mirroring functionality to be made in the same order on the mirror copy (target system
or partition).
Geographic mirroring allows for the production and mirrored copies to be at the same site for
high-availability protection in the event of server failure, and it is also possible to separate the
two systems geographically for disaster recovery protection in the event of a site-wide outage,
provided that the communication link between the two sites is enough for your
synchronization timing policy.
Geographic mirroring happens on the page level of storage management. Therefore, the size
of individual disks and the disk protection used on the production IASP can differ from what is
used on the backup IASP. The overall size of the two IASPs should be about the same on
both systems though.
Important: If both disk pools (source and target) do not have the same disk capacity
available, when the mirrored copy reaches 100%, geographic mirroring is suspended. You
can still add data to your production IASP, but you lose your high availability. If, however,
the production copy reaches 100%, you are no longer able to add data to that IASP, and
applications trying to do so come to a stop. An IASP can never overflow to the system
ASP, as that would compromise your high-availability environment.
Mirroring option
You can specify how to replicate from your production data in IASP to back up IASP.
Mirroring options of geographic mirroring can be specified as ASP session attributes, which
are a combination of two parameters:
Transmission delivery
Mirroring mode
Table 5-1 shows you the combination of these parameters that you can specify.
*SYNC *ASYNC
Figure 5-2 shows the DSPASPSSN command screen as configuring synchronous geographic
mirroring with synchronous mirroring mode.
…
Figure 5-4 Display ASP session command: Asynchronous geographic mode
Synchronization
When geographic mirroring is resumed after suspend or detach, the mirror copy will be
resynchronized with the production copy. The production copy can function normally during
synchronization, but performance might be negatively affected. During synchronization, the
contents of the mirror copy are unusable, and it cannot become the production copy. If the
independent disk pool is made unavailable during the synchronization process,
synchronization resumes where it left off when the independent disk pool is made available
again. Message CPI095D is sent to the QSYSOPR message queue every 15 minutes to
indicate progression of the synchronization.
Message CPI095D indicates what types of synchronization happened. Figure 5-5 shows
the message.
…
The synchronization is of type 2. The synchronization types and their meanings
are as follows:
1 - The synchronization being performed is a synchronization of tracked
changes.
2 - The synchronization being performed is a synchronization of all data.
…
Figure 5-5 Message CPI095D
Two parameters can be used to better manage IASP copies synchronization and application
performances when geographic mirroring is used:
Synchronization priority
When you set the attributes for geographic mirroring, you can set the synchronization
priority. You can select synchronization priority as high, medium, or low. If synchronization
priority is set high, the system uses more resources for synchronization, which results in a
sooner completion time. The mirror copy is eligible to become a production copy faster, so
you are protected sooner. However, high priority can cause degradation to your
application. We recommend that you try high priority first, so you are protected as soon as
possible. If the degradation to your application performance is not tolerable, then lower the
priority. Be aware that you need to vary off the IASP to perform this change.
Suspend timeout
In addition to synchronization priority, you can also set the suspend timeout. The
suspend timeout specifies how long your application can wait when geographic mirroring
cannot be performed. When an error, such as a failure of the communication link, prevents
geographic mirroring from occurring, the source system waits and retries for the
specified suspend timeout before suspending geographic mirroring, which allows your
application to continue.
Copy Descriptions
Bottom
Press Enter to continue
Tracking space
The tracking space was introduced with IBM i 5.4. It enables geographic mirroring to track
changed pages while in suspended status. With tracked changes we can avoid full
resynchronization after a resume in many cases, therefore minimizing exposed times where
you do not have a valid mirror copy. Tracking space gets configured when you configure
geographic mirroring or change geographic mirroring attributes via CHGASPSSN.
Tracking space is allocated inside of the independent ASPs. The more tracking space that
you specify, the more changes the system can track. The amount of space for tracking can be
defined by the user up to 1% of the total independent ASP capacity. When we specify tracking
space size, we can specify a percentage of total usable tracking space size. If you specify
100%, you have 1% of your total independent ASP capacity as tracking space size. For
example, if you have an independent ASP with 100 GB, you can have a maximum of 1 GB of
storage space as the tracking space, and if you specify the tracking space parameter as 50%,
you have 500 MB of storage space as tracking space. Be aware that this tracking space does
not contain any changed data. It just holds information about what pages in the IASP have
been changed.
When geographic mirroring is active in synchronous mode (Figure 5-7), the write on disk
operation waits until the operation is complete to the disk (actually the data write to the IOA
cache) on both the source (acknowledgement operation #4) and target systems
(acknowledgement operation #3) before sending the acknowledgment to the storage
management function of the operating system of the production copy. See the operations
numbered 1 - 4 with green arrows in Figure 5-7.
The mirror copy is always eligible to become the production copy, because the order of writes
is preserved on the mirror copy. If you are planning to use geographic mirroring as helocal HA
solution, we recommend trying synchronous mode first. If your performance remains
acceptable, continue to use synchronous geographic mirroring.
Site A Site B
Production Copy Mirror Copy
Main 2
Memory 3
4 1 TCP/IP Network
IOA Cache
IOA Cache
IASP
source
IASP
target
In IBM i 7.1, asynchronous mirroring mode can activate to specify that transmission delivery
is synchronous and mirroring mode is asynchronous.
Site A
Production Copy Site B
Mirror Copy
2
Main
Memory 3
4 1 Main
TCP/IP Network Memory
IOA Cache 5
IOA Cache
IASP
source
IASP
target
In this mode, the pending updates must be completed before the mirror copy can become the
production copy. This means that while you might see a slightly better performance during
normal operation, your switchover or failover times might be slightly longer because changes
to the backup IASP might still reside in the main memory of your backup system. They must
be written to disk before the IASP can be varied on.
Important: Because this mode still waits for the write to cross to the target system, it is not
truly asynchronous. We recommend this for situations in which the source and target
systems are less than 30 km apart.
Asynchronous transmission delivery, which also requires the asynchronous mirroring mode,
works by duplicating any changed IASP disk pages in the *BASE memory pool on the source
system and sending them asynchronously while preserving the write-order to the target
system in operation #3 in Figure 5-9. Therefore, at any given time, the data on the target
system (though not up-to-date) still represents a so-called crash-consistent copy of the
source system.
Site A
Production Copy Site B
Mirror Copy
3
Main
Memory 5
2 1 Main
TCP/IP Network Memory
IOA Cache 4
IOA Cache
IASP
source
IASP
target
The asynchronous geographic mirroring option has potential performance impacts to system
resources, such as processor and memory because communication lines with longer latency
times might tie up additional memory resources for keeping their changed data.
Bottom
F3=Exit F4=Prompt F5=Refresh F10=Additional parameters F12=Cancel
F13=How to use this display F24=More keys
Figure 5-10 Example of Change ASP session command (change session attributes)
For more information about performance of geographic mirroring see 8.3.1, “Geographic
mirroring” on page 144.
Source Target
LPAR-1 LPAR-1
LPAR-1
IASP
*SYSBAS
Tape backup
Adding replication to a remote site server provides additional levels of protection, including
disaster recovery or off-line tape backups. This combination can be achieved using internal
disks because PowerHA supports geographic mirroring with internal disks, and the replication
can be either synchronous or asynchronous.
Note: Geographic mirroring must first be suspended before IASP can be backed up to
tape. PowerHA tracks changes and requires time for partial resynch when backup is
complete, and hence affects your recovery point objective (RPO).
LPAR/ Node 3
Node 1
Switched
disk
Site A Site B
For additional information, especially when performance considerations apply, refer to IBM
System Storage DS8000: Architecture and Implementation, SG24-8886.
Array
.. … … 24 … … .. Site
.. … … 24 … … ..
Loop 1 Loop 2
Figure 6-1 An array site
All disk drives in an array site have the same capacity and the same speed.
6.1.3 Array
DS8000 logical storage configuration starts at this point. An array is a group of eight disk
drives created from an array site, defining the disk protection that you want to use. Available
disks protections are RAID10, RAID5, and RAID6. When creating the array, specify the array
site to use and the protection to apply. At this time, the DS8000 selects a spare disk to put in
the array, if needed according to sparing algorithm. None, one, or two spare disks per array
can exist.
Note: Spare disks, while they remain spare, are not a member of any RAID parity set
even if they are a member of the array. They become a member of the RAID parity set if
they replace a failing drive. These operations are automatically handled by the DS8000
when needed.
Array
Site D1 D7 D13 ...
D2 D8 D14 ...
D3 D9 D15 ...
D4 D10 D16 ...
D5 D11 P ...
D6 P D17 ...
Creation of
an array P D12 D18 ...
Data
Data
Data
Data RAID
Data Array Spare
Data
Parity
Spare
6.1.4 Rank
The next configuration step is to dedicate the array to one of the two disk formats provided by
the DS8000. It supports, at the same time, connections from both System z® servers and
Windows/Linux/Unix and IBM i operating systems, but disk formats are not the same for both
types. System z servers make use of count key data (CKD) format, while all the others make
use of fixed block (FB) format.
The rank defines the way that the entire array is formatted for CKD usage or FB usage. At
rank creation time, the array is also divided into extents. One extent is the smallest disk space
that can be allocated to DS8000 host operating systems.
For FB usage, one extent is equivalent to 1 GB (more precisely GiB, being equal to
230 bytes).
Figure 6-3 shows an example of a fixed block rank with 1 GB extents. In Figure 6-3, one
square represents one extent. The DS8000 distributes every extent to all disks members of
the array.
Data
Data D1 D7 D13 ...
D2 D8 D14 ...
Data
D3 D9 D15 ...
Data RAID D4 D10 D16 ...
Data Array D5 D11 P ...
Data D6 P D17 ...
Parity P D12 D18 ...
Spare
Creation
of a Rank
FB Rank
1 GB 1 GB 1 GB 1 GB
of 1 GB
FB FB FB FB extents
Note: From a DS8000 standpoint, the IBM i operating system and Windows/Unix/Linux
operating systems are members of the same community, the fixed block community. This
means that, for example, on the same rank there can coexist extents used by an IBM i
operating system and others used by a Windows operating system. You have to make
sure that you understand possible performance impacts with such a configuration.
Note: Server affinity applies only during normal operations. If a DS8000 server fails, the
other is able to manage extent pools previously managed by the failing server.
Server1
1 GB 1 GB 1 GB 1 GB
FB FB FB FB
1 GB 1 GB 1 GB 1 GB 1 GB 1 GB 1 GB 1 GB
FB FB FB FB FB FB FB FB
1 GB 1 GB 1 GB 1 GB 1 GB 1 GB 1 GB 1 GB
FB FB FB FB FB FB FB FB
Note: Performance considerations apply whether you decide to assign several ranks or a
single rank to an extent pool.
6.1.6 Volumes
Until this step, the DS8000 is not yet ready for providing any storage resources to host
operating systems. The next step is to create volumes, or logical volumes, by using a specific
extent pool. This one provides the volume, at creation time, with the desired number of
extents, located on the associated ranks. The volume does not take care of the extents’
physical location. Volume creation commands are not the same for CKD extent pools and for
FB extent pools because of parameter differences.
Figure 6-5 shows the process of creating a 2.9 GB FB volume in an extent pool spanned on
two ranks. Three extents (so 3 GB) remain available in the extent pool. Other extents are in
use by other LUNs. Creating the 2.9 GB volume succeeds but allocates the three remaining
extents, leaving 100 MB unused.
1 GB
Rank-a 1 GB 1 GB 1 GB
used
3 GB LUN 2.9 GB LUN
created
1 GB 1 GB
Rank-b Used Used
used used 100 MB unused
IBM i volumes are fixed block volumes, so they are also composed of a number of 1 GB
extents. However, for DS8000 native or NPIV attachment, the IBM i operating system
supports only certain fixed volumes sizes that are the same as supported physical disks
drives. For example, supported volume sizes are 8.58 GB, 17.54 GB, 35.16 GB, 70.54GB,
and so on. The other specific item for IBM i volumes is the disk protection that the operating
system wants to achieve. If you want to use host-based IBM i mirroring, IBM i must see
unprotected volumes. Otherwise it will be not possible to define and configure the mirrioring.
The protection flag provided by the DS8000 for an IBM i volume is only a way to make IBM i
believe that the volume is not protected. In reality, all volumes are RAID protected by the
DS8000 storage system, as discussed in 6.1.3, “Array” on page 84.
The DS8000 provides the IBM i operating system with a specific serial number for every
volume, just like physical disks drives have.
The DS8000 model determines the first two digits of the serial number (for example, DS8100,
DS8300, DS8700 report 50). They are the same for all volumes. The next four digits are the
volume ID. The last three digits are known as the OS/400 serial number suffix, which is set up
at the DS8000 storage image level. This is the same for all volumes, and it points you to the
appropriate DS8000, if your installation does not have a unique one.
Location: U8233.E8B.10001AP-V6-C60-T1-W50050763070102B8-L40EE409000000000
Logical address:
SPD bus:
System bus 255
System board 128
More...
Press Enter to continue.
An n-to-n relationship exists between volumes and volumes groups. However, if a volume
belongs to more than one volume’s group, it means that it is potentially accessed by several
hosts. In this case, the host operating system is responsible for ensuring data integrity. he
IBM i operating system does not support sharing a volume with another operating system.
Figure 6-9 shows an example of host connection relationships to volume groups. IBM i
partition number 1 has two Fibre Channel adapters, with WWPN 1 and WWPN 2. IBM i
partition number 2 has two Fibre Channel adapters, with WWPN 3 and WWPN 4. Inside the
DS8000, WWPN 1 and WWPN 2 are assigned the same volume group, volume group 1, and
WWPN 3 and WWPN 4 are assigned the same volume group, volume group 2. With this
configuration, IBM i partition 1 makes use of volumes included in volume group 1 and IBM i
partition 2 makes use of volumes included in volume group 2. Both the partitions have a dual
path to the volumes due to dual Fibre Channel attachments and assignment to the same
volume group.
WWPN 2
WWPN 4
Volume Volume
Group 1 Group 2
for Partition 1 for Partition 2
LSS play a lest important role for FB volumes than they do for CKD volumes. For more
information on LSS in a CKD context, refer to IBM System Storage DS8000: Architecture and
Implementation, SG24-8886. In FB context, LSS are important for some of Copy Services
usage which are detailed in next sections.
4 Write acknowledge
Server
write
1
2
Write to secondary
LUN or LUN or
volume 3 volume
Write complete
Primary acknowledgment Secondary
(source) (target)
Note: For all IBM i volumes involved in a Metro Mirror relationship, both source and
target volumes must have the same type-model. They must have the same capacity
and the same protection flag.
Note: A PPRC path must exist from the source to the target site before creating the Metro
Mirror pair.
Note: After terminating the Metro Mirror pair, you can delete the related PPRC path,
depending of path usage for other pairs.
Failback is related to the actions that you take to come back to the nominal situation, with
production running on initial source volumes. Failback operations can start after the initial
production site failure is fixed or maintenance is finished.
On the backup site, failover function performs three steps in a single task at the volume level:
1. Terminate the original Metro Mirror relationship.
2. Establish a new relationship in the opposite direction, from the initial target to the initial
source (from the backup site to the production site).
3. Before any update to the new source volume, suspend the new Metro Mirror relationship.
After these steps, the state of the original source volume is preserved (depending on the
failure, nobody might be able to connect to the original source DS8000), and the state of
When we are ready to switch back to the nominal situation, we can start failback. Still on the
backup site, the failback function performs various actions depending on the original source
volume state. Failback also works at the volume level:
If the original source volume is no longer in any Metro Mirror relationship, a new Metro
Mirror relationship between the original target volume and the original source volume is
established, and all data is copied back from the backup volume to the production volume.
If the original source volume is still a member of the appropriate Metro Mirror relationship
and has no updated tracks, only updated data on the original target volume is copied back
to the original source volume.
If the original source volume is still a member of the appropriate Metro Mirror relationship
and has some updated tracks, updated data on the original target volume and the same
tracks as the ones that were updated on the original source volume are copied back to the
original source volume.
After the failback, the Metro Mirror direction is from the original target volume to the original
source volume, and both volumes are synchronized.
The last operation to run to recover the original environment is an another failover and
failback to perform, this time on the original production site. Figure 6-11 summarizes all these
tasks.
For bandwidth and redundancy purposes, a path can have up to eight links. The DS8000
handles failures and balances the workload across available paths. As shown on Figure 6-12,
a link can handle several paths.
LSS 0 LSS 0
LSS 1 LSS 1
Physical
LSS 2 Fibre Channel link LSS 2
LSS 3 LSS 3
.. ..
LSS 08 256 logical paths LSS 08
.. per FCP link ..
LSS nn LSS nn
Tip: Do not forget to create a path in both directions, from production to backup DS8000
and from backup to production DS8000. Otherwise, failback operation is not possible.
Paths from backup to production can use any of the available links. There is no constraint
to use the same paths from production to backup.
Note: Although a DS8000 IO port can handle Metro Mirror traffic and host connections at
the same time, for performance reasons, we recommend not mixing them. We prefer
dedicating I/O ports to replication traffic or host connections.
Classical installation consists of a local IBM i partition and a remote IBM i partition. Both are
in a cluster, and each partition makes use of a DS8000 external storage for IASP disks. There
is no mandatory rule about SYSBAS disks. They can be internal drives or external storage.
Note: Metro Mirror relationships apply only to IASP volumes when used with IBM
PowerHA SystemMirror for i.
Figure 6-13 shows a setup overview of a Metro Mirror installation for an IASP. Both IBM i
partitions are members of a cluster, and Metro Mirror replication is in place for IASP volumes.
IBM i Cluster
DS
Metro
Mirror
IASP IASP
When a switch occurs, either for a planned outage or an unplanned one, the target IASP
volumes become available due to Metro Mirror failover operations, and the remote partition
can use them for production activity.
When coming back to a normal situation is possible, failback operations can start, and the
original source partition can start production activity.
As you see in 13.1.1, “Configuring IBM i DS8000 Metro Mirror (GUI and CL commands)” on
page 264, IBM PowerHA SystemMirror for i handles all setup, switch, and switch back tasks
but one. It is not able to establish a Metro Mirror relationship between source and target
volumes. This DS8000 setup is done either through DSCLI or the Storage Manager GUI.
Basically, Global Mirror combines two DS8000 techniques, Global Copy and Flash Copy:
Global Copy itself is not covered in this book, but it is almost exactly the same technique
as Metro Mirror, with two main differences:
– The replication is asynchronous. There is some delay before writes are effective on the
target site. RPO is not zero in this case. It mainly depends on network bandwidth and
latency between source and target sites. Figure 6-14 illustrates how Global Copy
operates when a host requires a write operation:
i. Data is stored in the source DS8000 cache and Non Volatile Storage, to be later
destaged to the source volume.
ii. Write acknowledgement is sent by the source DS8000 to the host.
iii. At a later time (that is, in an asynchronous manner), the source DS8000 sends the
necessary data so that the updates are reflected on the target volumes. The
updates are grouped in batches for efficient transmission.
iv. The target DS8000 returns write complete to the source DS8000 when the updates
are written to the target DS8000 cache and NVS.
b Write acknowledge
Server
write
a
LUN or LUN or
volume d volume
Write to secondary
Primary Secondary
(non-synchronously)
(source) (target)
– There is no guarantee that the write sequence on the target site is the same as it is on
the source site. Therefore, data is not consistent on the target site. It becomes
consistent when the application or host using the source volumes is quiesced. So data
migration is the main usage of Global Copy.
Host
Write I/O
F la
A B sh
Co
Source Target py
Global Copy
Copy Pending Copy Pending
C
Tertiary
O S T
O SBM: Source bitmap B B
S TBM: Target bitmap M M
OOS: Out-of-sync bitmap
Local site Remote site
Figure 6-15 Global Mirror implementation
1 Serialize all 2 3
Perform
Global Copy Drain data from local to remote site
FlashCopy
source volumes
O
C A1 B1
O
R C1
Source S Target
Tertiary
O
C A2 B2
O
R C2
Source S Target
Tertiary
Local Remote
Figure 6-16 Global Mirror consistency group formation
There are several parameters used for Global Mirror tuning. See the section “Consistency
Group interval time” in IBM System Storage DS8000: Copy Services in Open Environments,
SG24-6788. It determines how long to wait before starting with the formation of a new
Global Copy
This operation establishes the copy relationship between the source (or local) volume and the
target (or remote) volume. Normally, those two volumes reside on two DS8000s, but it is
possible to use the same DS8000 for tests purposes. During the copy, the target volume gets
a target copy pending state. For Global Mirror purposes, the Global Copy command with
default parameters values can be used.
Note: A PPRC path must exist from the source to the target site before creating the Global
Copy pair.
FlashCopy
Before creating the FlashCopy relationship on the target site, we recommend waiting until the
end of the Global Copy initial copy. Because of the asynchronous nature of Global Copy, you
cannot rely on the status, which will always be copy pending. However, as soon as the initial
copy step is finished you will see the out of sync tracks indicator close to zero.
Specific FlashCopy settings (Target inhibit, Record, No copy, and Persistent (which is
automatically set with Record)) are to be specified.
Failover
Failover operations for Global Mirror involve more steps than for Metro Mirror. The very first
one is the same as it is to reverse the Global Copy direction, so to make the target volumes
(B volumes) as source suspended. But, those volumes are not consistent. We have to play
with FlashCopy target volumes (C volumes), which are consistent. Therefore, these are the
overall operations for a failover:
1. Global Copy failover: The situation becomes as shown in Figure 6-17.
Server for
GM recovery
Failover B to A
FlashCopy
Global Copy
B
B
Source
Source
copypending Suspended C
Tertiary
Server for
GM recovery
FRR FlashCopy
B B
GC Source
Source FC Source
copypending C
Suspended Tertiary
FC Target
Failback
When the source site becomes available, in a Metro Mirror environment, the first step is to run
a failback operation from the target B volumes to synchronize the content of volume A with
the content of volume B.
The second step is to make volume A source back and to start synchronizing B volumes with
A volumes.
After Global Copy and FlashCopy are working again correctly, the last operation is to restart
the Global Mirror session.
Note: Each time that a failover operation is done, applications can be quiesced before to
have consistent data with no difference between the source and target data.
As with Metro Mirror replication, classical installation consists of a local IBM i partition and a
remote IBM i partition. Both are in a cluster, and each partition makes use of a DS8000
Note: Global Mirror relationships apply only to IASP volumes when used with IBM
PowerHA SystemMirror for i.
Figure 6-19 shows a setup overview of a Global Mirror installation for an IASP. Both IBM i
partitions are members of a cluster, and Global Mirror replication is in place for IASP volumes.
IBM i Cluster
DS
Global
Mirror
IASP IASP
Flash-
Copy
IASP in GM
When a switch occurs, either in case of a planned outage or an unplanned one, the target
IASP volumes become available due to Global Mirror failover, and the remote partition can
use them for production activity.
When coming back to a normal situation is possible, failback operations can start, and the
original source partition can start back production activity.
As you see in 13.1.3, “Configuring IBM i DS8000 Global Mirror (CL commands)” on page 311,
IBM PowerHA SystemMirror for i handles all setup, switch, and switchback tasks but one. It is
not able to establish the Global Mirror configuration between source and target volumes. The
DS8000 setup is done either through DSCLI or Storage Manager GUI.
IBM i Cluster
DS8000
*SYSBAS *SYSBAS
IASP
Production HA
A typical implementation scenario for LUN-level switching is where multi-site replication using
Metro Mirror or Global Mirror is used for disaster recovery and protection against possible
storage subsystem outages, while additional LUN-level switching at the production site is
used for local high-availability protection, eliminating the requirement for a site-switch in case
of potential IBM i server outages.
Bottom
F3=Exit F4=Prompt F5=Refresh F12=Cancel F13=How to use this display
F24=More keys
Figure 6-21 IBM i ADDASPCPYD enhancement for LUN-level switching
An ASP session is not required for LUN-level switching, as there is no replication for the
IASP involved.
You can see a more detailed scenario in 13.1.4, “Configuring IBM i DS8000 LUN-level
switching” on page 317.
Note: Setting up an ASP copy description for LUN-level switching is only supported from
the green-screen interface.
For LUN-level switching, the backup node host connection on the DS8000 or DS6000
storage system must not have a volume group (VG) assigned. PowerHA automatically
unassigns the VG from the production node and assigns it to the backup node at site
switches or failovers.
6.5 FlashCopy
In this section we introduce the DS8000 FlashCopy feature when used in IBM i environments.
For complete information, see Part 3, “FlashCopy,” in IBM System Storage DS8000: Copy
Services in Open Environments, SG24-6788.
Typically, large applications such as databases have their data spread across several
volumes, and their volumes should all be FlashCopied at exactly the same point-in-time.
FlashCopy offers consistency groups, which allows multiple volumes to be FlashCopied at
exactly the same instance.
If all bits for the bitmap of the target are set to their initial values, it means that no data block
has been copied so far. The data in the target is not modified during setup of the bitmaps. At
this first step, the bitmap and the data look as illustrated in Figure 6-22.
The target volume in the following figures can be a normal volume or a virtual volume (space
efficient volume). In both cases the logic is the same.
t0 tx ty tz time
bitmap
1 1 1 1 1 1
data data
t0 t0 t0 t0 t0 t0
t0 tx ty tz time
read 1 read 2
data data
t0 tx t0 tz t0 t0 t0 t0
Figure 6-23 Reads from source and target volumes and writes to source volume
To identify whether the data of the physical track on the source volume needs to be copied to
the target volume, the bitmap is analyzed. If it identifies that the time-zero data is not available
on the target volume, then the data will be copied from source to target. If it states that the
time-zero data has already been copied to the target volume then no further action is taken.
It is possible to use the target volume immediately for reading data and also for writing data.
t0 ta tb time
write a write b
bitmap
0 0 1 0 1 1
data data
t0 tx t0 ty t0 t0 ta t0 tb
Only the classical FlashCopy allows a full copy. FlashCopy SE has no such function. But
remember, both features can coexist.
Nocopy option
If FlashCopy is established using the nocopy option, then the result will be as shown in
Figure 6-22 on page 108, Figure 6-23 on page 109, and Figure 6-24.
The relationship will last until it is explicitly withdrawn or until all data in the source volume has
been modified. Blocks for which no write occurred on the source or on the target will stay as
they were at the time when the FlashCopy was established. If the persistent FlashCopy
option was specified, the FlashCopy relationship must be withdrawn explicitly.
The IBM System Storage DS8000 series FlashCopy SE licensed feature allows creation of
space-efficient FlashCopy target volumes that can help to reduce the required physical
storage space for the FlashCopy target volumes. These volumes are typically needed only for
a limited time (such as for the duration of a backup to tape).
A space-efficient FlashCopy target volume has a virtual storage capacity reported to the host
matching the physical capacity of the fully provisioned FlashCopy source volume, but no
physical storage space is ever allocated. Physical storage space for space-efficient
FlashCopy target volumes is allocated in 64 KB track granularity. This is done on demand for
host write operations from a configured repository volume shared by all space-efficient
FlashCopy target volumes within the same DS8000 extent pool (Figure 6-25).
Non-provisioned
space-efficient volumes
(no space ever allocated)
Repository Volume
(over-provisioned,
e.g., 500 GB virtual &
100 GB real capacity)
From a user perspective, the PowerHA setup (not the DS8000 FlashCopy setup) for
space-efficient FlashCopy is identical to the setup for traditional FlashCopy with the no-copy
option. The reason for this is that PowerHA SystemMirror for i internally interrogates the
DS8000 to determine the type of FlashCopy relationship and makes sure that it uses the
corresponding correct DSCLI command syntax. The syntax check is done for either
traditional FlashCopy or FlashCopy SE when issuing mkflash and rmflash.
For further information about using IBM System Storage DS8000 FlashCopy SE with IBM i,
see IBM Redbooks publication IBM System Storage Copy Services and IBM i: A Guide to
Planning and Implementation, SG24-7103.
The Storwize V7000 is a modular storage system that includes the capability to virtualize
external SAN-attached storage in addition to its own internal storage. The V7000 is built upon
the IBM SAN Volume Controller (SVC) technology base using corresponding storage
concepts and management interfaces with generally the same command set. From a
hardware perspective, the V7000 IBM device type 2076 consists of two dual-active
controllers with 8 GB of cache each, eight 8 Gb FC ports plus four optical 10 Gb FCoE ports,
up to 24 disks in the controller enclosure, and up to nine SAS-chain attached expansion
enclosures with up to 24 disks each, for a supported maximum of 240 internal small form
factor (SFF) disk drives. With the release of V6.2 microcode, a V7000 clustering option was
introduced to increase the V7000 scalability by allowing you to combine two V7000 systems
into a V7000 cluster with two I/O groups.
VD 7
VD 1
VD 2
VD 3
VD 4
VD 5
VD 6
Volumes / Virtual disks
Virtual-to-physical Mapping SVC / V7000
High Perf Low Cost Storage Virtualization
Storage pools / MDisk groups
MD 1
MD 2
MD 3
MD 4
MD 5
MD 6
MD 7
MD 8
Managed disks
LUN 1
LUN 2
LUN 3
LUN 4
LUN 1
LUN 2
LUN 3
LUN 4
SCSI LUNs
RAID RAID
controller 1 controller 2 SAN Storage
All virtual disks seen by the application servers are mapped to the physical storage in the pools
using a virtualization map that is fully transparent to the servers. Using this virtualization map,
the SVC/V7000 can duplicate the server data to make instant copies or copies at a remote
site, mirror data onto multiple physical disks to improve availability, migrate data concurrently
between physical storage systems, or over allocate its physical storage to provide
thin-provisioned virtual disks to the application servers and save physical storage cost.
The SCSI LUNs created on the physical storage systems typically configured as a single LUN
per RAID array are discovered as MDisks by the SVC/V7000 and assigned by the storage
administrator to storage pools/MDisk groups usually based on common performance and
availability device characteristics. Up to 128 storage pools with up to 128 MDisks each and up
to 4096 MDisks per cluster from up to 64 storage systems are supported. At creation of the
storage pool or MDisk group, an extent size ranging from 16 MB - 8 GB has to be specified,
which is used by the SVC/V7000 to address the storage pool and determines the maximum
Chapter 7. Storwize V7000 and SAN Volume Controller Copy Services 115
managed storage capacity in the cluster (from 64 TB to 32 PB) but is not relevant for the
cluster’s storage performance itself. Virtual disks (VDisks) are created from the extents of a
storage pool with up to 2048 VDisks supported per I/O group, that is, cluster node pair.
VDisks are created as either image mode, sequential mode, or striped mode VDisks
(Figure 7-2). By default, VDisks are created in striped mode, meaning that the extents for the
VDisks are allocated in a round-robin fashion from the available MDisks within the MDisk
group. That is, the strip size is identical to the extent size. Sequential mode VDisks have their
extents allocated sequentially from the available MDisks, meaning that another MDisk is used
for VDisk extent allocation only if the previous MDisk is already fully allocated. Image mode
VDisks represent a one-to-one mapping between MDisks and VDisks and are typically used
for migrating storage to (or from) the SVC/V7000.
Image Mode:
A Virtual Disk = Physical LUN
Sequential Mode:
B Virtual Disk mapped sequentially to
a portion of a single managed disk
C
Striped Mode:
I/O Group A I/O Group B Virtual Disk striped
across multiple managed disks
MDG1 MDG3
MDG2
For host write I/O processing, both nodes of an I/O group are involved (Figure 7-3 on
page 117) such that the host sends its write I/O request down the preferred path to the
preferred node, and the SVC/V7000 preferred node stores the write data in its UPS
backed-up write cache and propagates a second copy of the write data via the SAN fabric to
the alternate (non-preferred) node in the I/O group. After the write data has been stored in
both nodes’ write cache, it is acknowledged by an I/O complete message to the host and
asynchronously destaged by the preferred node to the physical disks. In case one of the
nodes of an I/O group becomes unoperational, the remaining node goes into write-through
mode, meaning that because the write cache redundancy has been lost, the host write I/O is
not acknowledged until it has been written to the physical disks with corresponding host write
I/O performance implications.
1 I/O Group 1
V1 V2
Preferred Alternative
node node
2
UPS1 SVC node 1 2 SVC node 2 UPS2
Cache Cache
Because both nodes of an I/O group are responsible for I/O processing, it is important that the
host systems are always zoned to both nodes of an I/O group and have the SDDPCM
multi-pathing software for SVC/V7000 installed.
Regarding the I/O to the disk storage system, the SVC/V7000 uses an integrated multi-path
driver within its microcode to direct the I/O for a specific LUN to a single storage system port
using a round-robin algorithm to distribute the workload from different LUNs to different
storage system ports. All LUNs that are part of an MDisk group or storage pool (that is, that
are managed MDisks) must be available. Otherwise, the complete MDisk group is taken
offline, except for an unavailable image mode MDisk, which does not impact the availability of
other MDisks.
Due to these I/O processing characteristics of the SVC/V7000, certain considerations are
required to ensure a proper SAN switch zoning without compromising redundancy or
performance. See 8.1.5, “Storage considerations” on page 139, for further information about
IBM PowerVM Virtual I/O Server requirements for SVC/V7000 attachment.
1
In SVC/V7000 terminology, a copy refers to one instance of an entity like a volume, that is, a mirrored volume has
two copies.
Chapter 7. Storwize V7000 and SAN Volume Controller Copy Services 117
7.1.4 Copy Services
The SVC/V7000 offers a common platform and single point of control not only for regular
provisioning and management of heterogeneous storage but also for advanced functions,
such as Copy Services, that are enabled by the SVC/V7000 virtualization-layer also between
storage systems of different architectures or from different vendors. Copy Services functions
available for the SVC/V7000 are Metro Mirror for synchronous remote replication, Global
Mirror for asynchronous remote replication, and FlashCopy for point-in-time volume copies
that are each discussed in more detail in the following sections.
The SVC/V7000 Copy Services functions are licensed on a used versus installed capacity
basis with a single combined license for Metro Mirror and Global Mirror. FlashCopy is
included in the V7000 base license (5639-VM1), whereas a separate license for FlashCopy is
required for the SVC based on the virtual capacity only of the source volumes.
Both intra-cluster (that is, within the same cluster (and within the same I/O group only, rather
than used for testing or migration purposes)) and inter-cluster remote copy relationships (that
is, between two SVC or V7000 systems) are supported by the SVC/V7000.
For inter-cluster remote Copy Services, two SVC or V7000 clusters are connected to each
other over a Fibre Channel fabric. The maximum supported distance for remote Copy
Services is determined by the maximum supported round-trip latency of 80 ms. Up to 10 km
Fibre Channel distance is supported with long-wave SFPs. Extended distance support is
provided by using extenders, routers, or DWDM/CWDM hardware with up to seven hop
counts supported.
The source and target volumes for Copy Services relationships need to be of the same
virtual size.
Interoperability between SVC and V7000 Copy Services is available with the latest
SVC/V7000 V6.3 microcode and is also supported by 5799-HAS PRPQ PowerHA
SystemMirror for i. Allowing a V7000 system to participate in a SVC remote copy cluster
partnership requires its cluster layer property to be changed from storage to appliance.
For further details about SVC and V7000 refer to these IBM Redbooks publications:
Implementing the IBM System Storage SAN Volume Controller V6.1, SG24-7933
Implementing the IBM Storwize V7000, SG24-7938
Host
1 Write 4 Ack
3 Ack Write
Cache Cache
2 Mirror Write
Master Auxiliary
volume Metro Mirror Relationship volume
(primary (secondary
role) role)
The role of a master or auxiliary volume is either primary or secondary, depending on the
direction or failover state of the current remote copy relationship. Up to 2048 remote copy
relationships are supported in a two-node SVC/V7000 cluster.
With SVC/V7000, establishing a Copy Services relationship is done in two phases of creating
the relationship first before starting it in a second step. This is different from, for example, IBM
System Storage DS8000 Copy Services, for which establishing a Copy Services relationship
is done in a single step by creating the out-of-sync bitmaps and starting the relationship
automatically at its creation.
When creating the Metro Mirror relationship, the user can specify whether the auxiliary
volume is already in sync with the master volume, and the background copy process is then
skipped. The in-sync option (path 1a in Figure 7-5 on page 120) is intended to be used when
the volumes were created with the format option should not be used for IBM i volumes
because IBM i specially formats the volumes itself when they are configured (that is, added to
an IBM i ASP). Hence, using the SVC/V7000 format option at volume creation and therefore
the in-sync option when creating a Metro Mirror relationship does not make sense.
Chapter 7. Storwize V7000 and SAN Volume Controller Copy Services 119
systems to prevent performance impacts for the foreground host I/O. Additionally, there
is also a relationship bandwidth limit for the maximum background copy rate for each
remote copy relationship hat defaults to 25 MBps and is an attribute of the SVC/V7000
cluster configuration.
Figure 7-5 shows an overview of these states as they apply to a connected remote copy
relationship and the conditions that cause a state transition.
Create
1a Remote Mirror 1b
) Relationship (o u
s y nc t of
syn
(in c)
Consistent Inconsistent
Stopped Stopped
Fo
r
Stop ( o u ce d Stop
t o Sta
fs r
2a Start or yn t 2b Start or
Error c) Error
Stop
4b enable
access 3
Consistent Inconsistent
Synchronized Background copy complete Copying
Start
5a
(in sync)
art
Stop
d St c)
e n
4a enable rc sy
access Fo t of
u
(o
Idling
In addition to these states that are valid for a connected remote copy relationship (that is, one
where the primary system is still able to communicate with the secondary system), there is
also a disconnected state of remote copy relationships where the primary system can no
longer communicate with the secondary system. When the clusters can communicate again,
the relationships automatically become connected again.
Chapter 7. Storwize V7000 and SAN Volume Controller Copy Services 121
If the relationship or consistency group becomes disconnected, the primary volumes
transition to inconsistent disconnected. The master side transitions to idling disconnected.
The SVC/V7000 logs informational events like remote copy relationship changes, loss of
synchronization, or remote cluster communication errors in an error log for which SNMP
traps, e-mail notification, or syslog server messages can be configured to trigger either
automation or alert the user for manual intervention.
For Metro Mirror and Global Mirror remote copy relationships, maintaining the correct write
order processing requires that in case of an error event causing loss of replication for only a
subset of remote copy relationships of an application servers, remote write update processing
for the non-affected remote copy relationships is automatically stopped to ensure application
server data consistency at the secondary site.
For FlashCopy point-in-time volume copies, maintaining consistency and correct write order
processing requires that write I/O to all volumes of an application server in a consistency
group is temporarily put on hold until all FlashCopy volume relationships for the consistency
group have been started. The storage system depends on the concept of dependent writes
having been implemented in the application logic to ensure consistency across multiple
volumes in a consistency group (for example, that a journal is updated with the intended
database update before the database itself is actually updated). This application logic write
dependency ensures that when a SCSI queue full status is set as part of the consistency
group formation for a volume, further dependent application writes are put on hold by the
application so that the storage system can proceed setting SCSI queue full status for all
remaining volumes and therefore guarantee dependent write data consistency for all volumes
in the consistency group. This write dependency concept still applies for IBM i with its
single-level storage architecture, as IBM i SLIC storage management would hold off all I/O to
a disk unit in an SCSI queue full condition but would not stop the I/O to other disk units that
are still available for I/O operations.
PowerHA SystemMirror for i with the 5799-HAS PRPQ inherently uses consistency groups
for SVC/V7000 FlashCopy relationships and requires them to be configured for Metro Mirror
or Global Mirror relationships. Due to the IBM i single-level storage architecture, which stripes
the data across all disk units of an ASP, consistency groups should be defined on an IASP
group level.
Note: Stand-alone volume copy relationships and consistency groups share a common
configuration and state model. That is, all volume copy relationships in a consistency
group that is not empty have the same state as the consistency group.
Host
1 Write 2 Ack
4 Ack Write
Cache Cache
3 Mirror Write
Master Auxiliary
volume Global Mirror Relationship volume
(primary (secondary
role) role)
Though the data is sent asynchronously from the primary to the secondary SVC/V7000, the
write ordering is maintained by sequence numbers assigned to acknowledged host write I/Os
and with the secondary applying writes in order by their sequence number. Consistency of
the data at the remote site is maintained at all times. However, during a failure condition the
data at the remote site might be missing recent updates that have not been sent or that were
in-flight when a replication failure occurred, so using journaling to allow for proper crash
consistent data recovery is of key importance, just like we generally recommend ensuring
crash consistent data recovery even when not using remote replication to be able to recover
from a loss of transient modified data in IBM i main store after a potential sudden server crash
due to a severe failure condition.
Global Mirror volume relationship states and transitions are identical to those for Metro Mirror,
as described previously in 7.2.2, “Remote copy relationship states” on page 120.
A log file is utilized by the SVC/V7000 for Global Mirror to maintain write ordering and help
prevent host write I/O performance impacts when the host writes to a disk sector that is either
currently in the process of being transmitted or due to bandwidth limits is still waiting to be
transmitted to the remote site. The SVC/V7000 also makes use of shared sequence numbers
to aggregate multiple concurrent (and dependent) write I/Os to minimize its Global Mirror
processing overhead.
Chapter 7. Storwize V7000 and SAN Volume Controller Copy Services 123
write response times above 5 ms is tolerated. If this tolerated duration of degraded performance
where the SVC/V7000 needs to hold off writes to the primary volumes with the effect of
synchronous replication-like degraded performance is exceeded, it stops the most busy active
Global Mirror relationship consistency group to help protect the application server’s write I/O
performance and logs an event with error code 1920. The link tolerance can be disabled by the
user setting its value to 0. However, this provides no protection for the application server’s write
I/O performance anymore in cases where there is congestion on either the replication link or the
secondary storage system. While the link tolerance setting allows you to define a period of
accepted performance degradation, it is still important to properly size the remote copy replication
bandwidth for the peak write I/O throughput and possible re-sync workload, and to help prevent
longer production workload performance impacts.
7.4 FlashCopy
The SVC/V7000 FlashCopy function provides the capability to perform a point-in-time copy of
one or more volumes (VDisks). In contrast to the remote copy functions of Metro Mirror and
Global Mirror, which are intended primarily for disaster recovery and high-availability
purposes, FlashCopy is typically used for online backup or creating a clone of a system or
IASP for development, testing, reporting, or data mining purposes.
PowerHA SystemMirror i with the 5799-HAS PRPQ supports these SVC/V7000 FlashCopy
functions, which are discussed in more detail below:
FlashCopy no-copy and background copy
Thin-provisioned (space-efficient) FlashCopy targets
Incremental FlashCopy
Reverse FlashCopy
Multi-target FlashCopy (by using separate ASP copy descriptions for each target)
Reads
FlashCopy Bitmap
Source Target 1 1 1 1 1 1 1 1 1 1
1 1 1 1 1 1 1 1 1 1
1 1 1 1 1 1 1 1 1 1
1 0 1 1 1 1 1 1 1 1
1 1 1 1 1 1 1 1 1 1
1 1 1 1 1 1 0 1 1 1
1 1 1 1 1 1 1 1 1 1
1 1 1 1 1 1 1 1 1 1
Writes
FlashCopy Bitmap
Source Target 1 1 1 1 1 1 1 1 1 1
1 1 1 1 1 1 1 1 1 1
1 1 1 1 1 1 1 1 1 1
1 0 1 1 1 1 1 1 1 1
1 1 1 1 1 1 1 1 1 1
1 1 1 1 1 1 0 1 1 1
1 1 1 1 1 1 1 1 1 1
1 1 1 1 1 1 1 1 1 1
While a fixed grain size of 256 KB is used for remote mirror volume relationships for
FlashCopy, the user can choose from the default grain size of 256 KB or alternatively from the
smaller grain size of 64 KB as the granularity for tracking and managing out-of-sync data of a
FlashCopy relationship.
Chapter 7. Storwize V7000 and SAN Volume Controller Copy Services 125
The concept of consistency groups to ensure that dependent write data consistency across
multiple volume copy relationships applies to FlashCopy (see 7.2.3, “Consistency groups” on
page 122).
When creating a FlashCopy relationship, the user can specify a desired copy rate for the
background copy process, which can either be 0 (meaning that a FlashCopy no-copy
relationship without a background copy is established) or any value from 1 - 100, which
translates to the desired background copy throughputs (Table 7-1).
11 - 20 256 KBps 1 4
21 - 30 512 KBps 2 8
31 - 40 1 MBps 4 16
41 - 50a 2 MBps 8 32
51 - 60 4 MBps 16 64
61 - 70 8 MBps 32 128
71 - 80 16 MBps 64 256
Chapter 7. Storwize V7000 and SAN Volume Controller Copy Services 127
consistent state of the source volume on disk in preparation for starting the
relationship. Any read or write data associated with the target volume is
discarded from the cache.
Prepared Relationship is ready to be started with the target volume having been
placed in offline state. Write performance for the source volume can be
degraded, as it is in write-through mode.
Note: Using thin-provisioned SVC/V7000 volumes for IBM i is only supported for
FlashCopy target volumes, not for volumes directly assigned to an IBM i host.
Using thin-provisioned volumes also does not make sense for remote copy secondary
volumes, which become fully allocated at initial synchronization. Similarly, using
thin-provisioned volumes for FlashCopy targets only makes sense for FlashCopy no-copy
relationships that are used for a limited duration and therefore have limited changes only.
For optimal performance of thin-provisioned FlashCopy, the grain size for the thin-provisioned
volume used as the FlashCopy target should match the grain size of the FlashCopy
relationship. Additionally, for space-efficiency reasons, to help minimize physical storage
allocations on the thin-provisioned target, consider also using the small grain size of 64 KB.
Original t0
Source Target
multi-target Volume B
Volume A
relationships
t0+1 Older target
depends upon
newer target
Target
Volume C
Later recovery
2 Reverse
by reverse FlashCopy
FlashCopy operation
Source Target
Volume A Volume B
1 Optional copy of
original source OR
Target
Volume C
Target
Volume D
A key advantage of the SVC/V7000 reverse FlashCopy function is that it does not destroy the
original source and target relationship, so any processes using the target (such as tape
backup jobs) can continue to run uninterrupted. It does not require a possible background
copy process to have completed, and regardless of whether the initial FlashCopy relationship
is incremental, the reverse FlashCopy function only copies data from the original target to the
source for grains that were modified on the source or target.
Consistency groups cannot contain more than one FlashCopy relationship with the same
target volume, so they need to be reversed by creating a set of new reverse FlashCopy
relationships and adding them to the new reverse consistency group.
The SVC/V7000 performs the copy on write processing for multi-target relationships in a way
that data is not copied to all targets, but only once from the source volume to the newest target
so that older targets refer to newer targets first before referring to the source. Newer targets
together with the source can therefore be regarded as composite source for older targets.
Chapter 7. Storwize V7000 and SAN Volume Controller Copy Services 129
130 PowerHA SystemMirror for IBM i Cookbook
8
In this chapter, we discuss important planning considerations and all the prerequisites that
you need to use IBM PowerHA SystemMirror for i in your environment.
The licensed program 5770-HAS must be licensed for all processor cores in the partitions
(nodes) that you want to be part of your cluster.
There are two editions of PowerHA System Mirror for i, the Standard edition and the
Enterprise edition (Table 8-1). The standard edition, which is PID 5770-HAS, is targeted at a
data center HA solution. The Enterprise edition, which is a feature of 5770-HAS, adds support
for a multi-site HA and DR solution. There is no charge to upgrade from PowerHA 6.1 to 7.1
Standard or Enterprise Edition.
Integrated heartbeat
Application monitoring
LUN-level switching
The pricing of PowerHA SystemMirror 7.1 for i is based on a per-core basis and is broken
down among small-tier, medium-tier, and large-tier Power server for each edition.
You can order a Power System Capacity BackUp (CBU) model for your disaster recovery
and high availability when ordering a Power System or doing a model upgrade. The terms for
the CBU for i allow a primary system processor's optional i license entitlements and 5250
Enterprise Enablement license entitlements (or optional user entitlements in the case of the
Power 520 and 720 CBU) to be temporarily transferred to the CBU Edition when the primary
system processor cores are not in concurrent use. For an IBM i environment, you must
register the order for a new CBU at IBM Capacity Backup for Power Systems site:
http://www-03.ibm.com/systems/power/hardware/cbu/
Production CBU
Partition #1
Unused
Partition #2
Active standby
CBU Cores
Partition #3
Active standby
CBU Cores
This PRPQ (5799-HAS) is available in the English language (2924) only but can be installed
on other language systems. You must install the PRPQ with the LNG parameter as 2924
when you install PRPQ with RSTLICPGM. PTF SI44148 for LPP 5770-HAS is a prerequisite for
the PRPQ.
When using external storage-based remote Copy Services together with IBM PowerHA
SystemMirror for i, make sure to use separate Fibre Channel adapters for SYSBAS and for
each IASP group. When using NPIV, you do not need separate physical adapters, but you
still must use separate virtual adapters for SYSBAS and each IASP group.
Less than 20 3
20 - 40 4
Greater than 40 5
Note: Every situation is going to be different. The rule of thumb above is presented as a
starting point. Your own requirements might vary.
Also, install the latest PTFs related with IBM PowerHA SystemMirror for i to all nodes in the
cluster. The latest known PTFs fix known problems and provide the enhanced benefits in an
IBM PowerHA environment. Periodically check and install the latest known PTF in
“Recommended Fixes for High Availability for Release 7.1" at the following IBM i Support
Recommended fixes website:
http://www-912.ibm.com/s_dir/slkbase.nsf/recommendedfixes
IBM PowerVM
IBM PowerVM is a virtualization technology for AIX®, IBM i, and Linux environments on IBM
POWER processor-based systems. The PowerVM Virtual I/O Server included in the
PowerVM provides a virtualization environment, in that the storage and network I/O
resources are shared. This feature is required for IBM i to attach to DS8000 via VSCSI or
NPIV and to SVC/V7000.
They provide logical partitioning (LPAR) technology by using either the Hardware
Management Console (HMC) or the Integrated Virtualization Manager (IVM), Dynamic LPAR
operations, Micro-Partitioning®, and Virtual I/O Server capabilities, and N_Port ID
Virtualization (NPIV). You can choose the editions of IBM PowerVM depending on what
features you want (Table 8-3).
Suspend/resume
NPIV
Thin Provisioning
Active Memory™
Sharing
The Virtual I/O Server owns the physical I/O resources, such as Ethernet and SCSI
(SAS)/FC adapters. It virtualizes those resources for its client LPARs to share them remotely
using the built-in hypervisor services. These client LPARs can be quickly created, typically
owning only physical memory and shares of processors without any physical disks or physical
Ethernet adapters.
Traditionally, IBM i has been supporting 520 bytes per sector storage. There are restrictions
to directly attach a common storage system to IBM i for this reason. However, VIOS brings to
he IBM i client partition the ability to use 512 bytes per storage sector by using the sector
conversion of IBM PowerVM Hypervisor, which converts IBM i traditional 8 x 520 byte sectors
into 9 x 512 byte sectors of a 4 KB memory page. Therefore, open storage volumes (or logical
units, LUNs) are physically attached to VIOS through a FC or a Serial-attached SCSI (SAS)
connection and then made available to IBM i (Figure 8-2).
FC Adapter
Device Driver IBM i Multipathing Device Driver
Multi-pathing Multi-pathing
SCSI
LUNs
Hdisk #1-n Hdisk
#1-n #1-n
FC Adapter
FC Adapter
POWER Hypervisor
Virtual SCSI
From IBM i 6.1.1, IBM i VSCSI client driver supports MPIO through two or more VIOS
partitions to a single set of LUNs. This multipath configuration allows a VIOS partition to fail or
be brought down for service without IBM i loosing access to the disk volumes as the other
VIOS partitions remain active (Figure 8-2 on page 137).
When setting up external storage with VIOS and VSCSI, there are several VIOS settings that
should be changed. The first is the fc_err_recov and dyntrk attributes for each fscsiX device
in VIOS partitions. To show the current fscsiX devices, use lspath.
Run the following command to set the fast fail and dynamic tracking attributes on
these devices:
chdev–attr fc_err_recov=fast_fail,dyntrk=yes –perm –dev fscsiX
R to restart VIOS after these changes. The fast fail is designed for a multipath environment
not to retry failed paths for a long time. In single path configuration, do not set fast_fail.
Note: In a dual VIOS environment using IBM i multipath, the SCSI reserve policy for each
LUN (or hdisk) on both VIOS LPARs must be set to no_reserve to allow disk sharing.
This change must be made prior to mapping the LUNs to IBM i, and it does not require a
restart of VIOS.
To show the current reserve policy settings, run the following command:
lsdev –dev hdiskX –attr reserve_policy
You can set the no_reserve attribute by running the following command:
chdev –dev hdiskX –attr reserve_policy=no_reserve
You need to install the SVC Subsystem Device Driver Path Control Module (SDDPCM) on
VIOS to enable the management of multiple paths to the SAN Volume Controller or V7000
virtual Disks (VDisks). For more information about the Multipath Subsystem Device Driver,
see the latest user's guides, available here:
https://www-304.ibm.com/support/docview.wss?dc=DA400&rs=540&uid=ssg1S7000303&conte
xt=ST52G7&cs=utf-8&lang=en&loc=en_US
WWPNs
aaaabbbb11112222
ccccdddd33334444
VFC VFC
Server Client
Adapter ID10 Adapter ID10
IBM i LUNs Mapped to
WWPN aaaabbbb11112222
POWER Hypervisor
For more information about connecting IBM i through VIOS to DS8000 using VSCSI and
NPIV, see DS8000 Copy Services for IBM i with VIOS, REDP-4584.
For more information about using DS CLI with IBM i, see Chapter 8, "Using DS CLI with
System i," in IBM i and IBM System Storage: A Guide to Implementing External Disks on
IBM i, SF24-7120.
Fibre Channel-attached LUNs are identified as the storage unit device type of 2107 on the
IBM i host system. You can specify 1 - 32 LUNs for each attachment to the IBM i Fibre
Channel adapter feature 2766, 2787, or 5760. Also, you can specify 1 - 64 LUNs for each
attachment to the IBM i Fibre Channel adapter feature 5749, 5774/5276, or 5735/5273.
SVC/V7000
Attaching SVC/V7000 to IBM i is possible using IBM PowerVM Virtual I/O Server (VIOS).
IBM i runs in its own LPAR as a client of the VIOS server and accesses the SVC/V7000
storage via Virtual SCSI (vSCSI).
For this attachment, the minimum requirements for IBM PowerHA SystemMirror for i are
SVC/V7000 Version 6.1.x and IBM VIOS 2.2.1. For a SVC-supported hardware list, device
driver, firmware, and recommended software level, see the following web page:
https://www-304.ibm.com/support/docview.wss?uid=ssg1S1003697#_CodeLevel
Table 8-5 lists the configuration maximums for VIOS supporting an IBM i client attached to SVC.
Volume (HDisk) 512 The maximum number of volumes that can be supported
by the SAN Volume Controller for a host running an IBM i
operating system (per host object).
Paths per volume 8 The maximum number of paths to each volume. The
suggested number of paths is 4.a
a. Subsystem device driver path-control module (SDDPCM) for AIX supports 16 paths per
volume, but the SAN Volume Controller supports a maximum of only eight paths for a
reasonable path-failover time.
IASP-A
Prod DS
Prod
System/LPAR
*SYSBAS
Global
Mirror
SAN
IASP-A IBM i Cluster & Device Domain
Global Copy
FlashCopy
D/R
System/LPAR
DR DS *SYSBAS
IASP-A
Flash Copy w/CG
After production workloads are moved to the recovery site, Global Copy must be used to
return to the primary site. This is a manual procedure that is not supported by IBM PowerHA
SystemMirror for i. For more detailed steps for Global Mirror switchback with an asymmetrical
configuration, see in “Using CL commands for an asymmetrical Global Mirror switchback after
a planned outage” on page 339. Because no disaster recovery capability would be provided
in the reverse direction, it is unlikely that in this type of configuration we would choose to run
for extended periods of time in the secondary location unless forced to by unavailability of the
primary site.
Because Global Mirror uses two copies of data in the secondary location, there would be
twice as many physical drives in this location as in the production location if the same size
drives were used. In some situations, it might be cost effective to use larger drives in the
secondary location. Spreading the production data over all these drives should provide
equivalent performance in a disaster situation while reducing the overall cost of the solution.
IASP-A
Prod DS
Prod
FlashCopy
System/LPAR
*SYSBAS
IASP-A
Flash Copy w/CG
Global
SAN
Mirror
D/R
FlashCopy
System/LPAR
*SYSBAS
DR DS
IASP-A
Flash Copy w/CG
With the standard FlashCopy, full copy is the default, while the nocopy option is the default
for FlashCopy SE.
To perform an incremental FlashCopy, you must first establish the FlashCopy relationship
with the change data recording and persistent FlashCopy options enabled. You can do the
incremental copy at any time, and you do not have to wait for the previous background copy
to complete.
Usually you can use the incremental FlashCopy to minimize the amount of data that must be
regularly copied and save the time for the physical copy of FlashCopy.
With reverse FlashCopy, the FlashCopy relationship can be reversed by copying over
modified tracks from the target volume to the source volume. For DS6000 and DS8000, the
background copy process must complete before you can reverse the order of the FlashCopy
relationship to its original source and target relationship. The change data recording is a
prerequisite for reverse restore.
The reverse restore function can only be used when a full copy relationship is completed.
Therefore, it is not possible with FlashCopy SE. PowerHA supports reverse FlashCopy only
to volumes that are not a part of a Metro Mirror or Global Mirror relationship.
In the following sections we discuss how to size for good geographic mirroring performance.
There are two separate aspects to consider when sizing for a geographic mirroring
environment. During the normal run time of the production environment, there will be some
overhead added by geographic mirroring as the IBM i operating system is sending disk writes
to the target system. The second aspect is the overhead and time required for
synchronization, when the target IASP is reconnected to the source IASP and changes are
pushed from the source to the target to make the two equivalent again.
CPU considerations
There is extra overhead on CPU and memory when doing geographic mirroring, and this has
to be considered for both the source and the target system. Geographic mirroring increases
the CPU load to the system processors on both the system owning the production copy of the
IASP and the system owning the mirror copy of the IASP. There must be sufficient excess
CPU capacity to handle this overhead, but there is no formula to calculate this exactly as it
depends on many factors in the environment and the configuration. This CPU usage is
needed for both systems to communicate and replicate data from the source IASP to the
target IASP.
You might require additional processors to increase CPU capacity. As a general rule, the
partitions that you are using to run geographic mirroring needs more than a partial processor.
In a minimal CPU configuration, you can potentially see 5 - 20% CPU overhead while running
geographic mirroring.
Regarding the backup system, be especially careful in sizing that system's processor. It
should not be a small percentage of your production system, because this might slow down
synchronization times considerably. If your backup system has fewer processors in
comparison to your production system and there are many write operations, CPU overhead
might be noticeable and affect performance.
Memory considerations
Geographic mirroring also requires extra memory in the machine pool. For optimal
performance of geographic mirroring, particularly during synchronization, increase your
machine pool size by at least the amount given by the following formula and then use
WRKSHRPOOL to set the machine pool size:
Extra machine pool size = 300 MB + (0.3 * number of disk arms in the IASP)
This extra memory is needed particularly during the synchronization process on the system
that owns the mirror copy of the IASP. However, you must add extra storage on every cluster
Important: The machine pool storage size must be large enough before starting the
resynchronization. Otherwise, increasing memory is not taken into account as soon as the
synchronization task is in progress, and the synchronization process can take longer.
If the system value QPFRADJ is equal to 2 or 3, then the system might make changes to the
storage pools automatically as needed. To prevent the performance adjuster function from
reducing the machine pool size take these steps:
1. Set the machine pool minimum size to the calculated amount (the current size plus
the extra size for geographic mirroring from the formula) by using the Work with
Shared Storage Pools (WRKSHRPOOL) command or the Change Shared Storage Pool
(CHGSHRPOOL) command.
2. Set the Automatically adjust memory pools and activity levels (QPFRADJ) system value to
zero, which prohibits the performance adjuster from changing the size of the machine pool.
Note: We recommend that you use WRKSHRPOOL for setting the machine pool size to the
calculated minimum. Disabling Performance Auto Adjuster can have other performance
implications in your environment.
Less than 20 3
20 - 40 4
Greater than 40 5
For example, if IASP contains 10 drives, then SYSBAS should have at least three. As another
example, if IASP contains 50 drives, then SYSBAS should have at least 10.
Note: Disk pool sizing is very application dependent, and the above guidelines are only
given as a starting point. You might find that fewer or more arms are required for your
application environment. Understanding performance monitoring for your application
environment is critical for sizing and capacity planning.
You will want to monitor the percent busy of the SYSBAS disk arms in your environment to
ensure that you have the appropriate number of arms. If it gets above 40% utilization, then
you must add more arms. Also, when possible, the disk assigned to the IASP should be
placed on a separate I/O adapter from the SYSBAS disk to reduce any potential contention. It
has also been found that IOA cache is very important and provides greater data integrity and
improved performance.
Communications lines
When you are implementing a PowerHA solution using geographic mirroring, plan for
adequate communication bandwidth so that the communications bandwidth does not become
a performance bottleneck in addition to system resources.
Geographic mirroring can be used for virtually any distance. However, only you can determine
the latency that is acceptable for your application. The type of networking equipment, the
quality of service, the distance between nodes, the number, and the characteristics of data
ports used can all affect the communications latency. As a result, these become additional
factors that can impact geographic mirroring performance.
To ensure better performance and availability, we recommend that you take the
following actions:
To provide consistent response time, geographic mirroring should have its own redundant
communications lines. Without dedicated communication lines, there might be contention
with other services or applications that utilize the same communication line. Geographic
mirroring supports up to four communication lines (data port lines), and a cluster heartbeat
can be configured for up to two lines. However, we recommend utilizing Virtual IP
addresses to provide redundancy to the cluster heartbeat.
It is important to know that a round-robin approach is used to send the data across the
lines. This implies that for best performance, when multiple dataport lines are configured,
they should have close to equivalent performance characteristics. If one slow line is
added, then this will gate the sending of the data to that line speed.
Geographic mirroring replication should also be run on a separate line from the cluster
heartbeating line (the line associated with each node in the cluster). If the same line is
used, during periods of heavy geographic mirroring traffic, heartbeating could fail, causing
Node 1 Node 2
PRODUCTION Copy MIRROR Copy
LAN 1
LAN 2
SYSBAS SYSBAS
Geographic
Mirroring
using multiple
Ethernet adapters
IASP and LANs IASP
source target
Site A Site B
Runtime environment
In this section we discuss the runtime environment.
Synchronous delivery and synchronous mode guarantee equivalent copies of the IASP on
the source and the target while geographic mirroring is active. It also provides the added
protection of a crash-consistent copy of the data in case of a target system failure, because all
writes will have been received into the disk subsystem.
Synchronous delivery and asynchronous mode can be beneficial for customers running with
a significantly slower disk subsystem on their target system. This allows the disk write on the
Asynchronous delivery is best for those environments in which the source and target are
separated by too long of a distance for acceptable synchronous response times, or for
scenarios where the bandwidth cannot support the peak write rate.
For synchronous delivery, the bandwidth of the communications lines must be able to keep
up with the peak write volume. If it cannot keep up, the writes will begin to stack up and
production performance will suffer.
For asynchronous delivery, the bandwidth of the lines must still keep up at least to the
average write volume. Because writes on the source are not waiting, it is acceptable for some
queuing to occur, but if the line cannot handle the average write volume, then geographic
mirroring will continue to get further and further behind.
It also is important to examine the variance of the write rate over time. If there is a large
variance between peak and average, then it might be advisable to size more for the peak.
Undersizing in this case affects the recovery point objective in the case of a source system
failure during the peak write rate.
T3 lines are a common aggregation of 28 T1 circuits that yield 44.736 Mbps total network
bandwidth or 5.5 MBps with a best effective throughput of 70%, which equals 3.9 MBps and a
planning number of 2 MBps.
The OC (the optical carrier fiber optic-based broadband network) speeds help you grow.
Synchronization
In this section we discuss synchronization.
If geographic mirroring is suspended without tracking, then full synchronization occurs. This
can be a lengthy process. If geographic mirroring is suspended with the tracking option,
PowerHA will track changes up to the tracking space limit specified on the ASP session.
When mirroring is resumed, the production and mirror copies are synchronized concurrently
with performing geographic mirroring.
Tracking is available on both the source side and the target side. Target side tracking greatly
reduces the need for a full synchronization. Usually a full synchronization is only required
when either the source or target IASP does not vary off normally, such as from a crash or an
abnormal vary-off.
Tracking space
Tracking space is a reserved area within the IASP where the system tracks changed pages
while in suspended status, and the changes need to be synchronized when resuming
mirroring. Tracking space is needed only when the target copy of the IASP is suspended,
detached, or resuming. The changes themselves are not contained within the tracking space,
only a space-efficient indication of which pages require changes. The amount of tracking
space allocated can be defined by the user. The maximum is 1% of the total space within the
IASP. Using CHGASPSSN, a user can set the percentage of that 1%. For example, setting the
field to 10% means that the tracking space would be 10% of 1% or .1% of the total IASP size.
These parameters can be viewed using DSPASPSSN. The tracking space allocated is the
percentage of the maximum (it would show 10% in the above example) and the tracking
space used is the percentage of the available tracking space being used.
Note: If tracking space is exhausted (it reaches 100%), then no more changes can be
tracked, and when you resume geographic mirroring, a full synchronization is required.
Monitoring synchronization
To track how much data is left to be synchronized, DSPASPSSN can be used on the source
system. On the second screen, there are fields for Total data out of synch and Percent
complete. These fields will display the megabytes of data that need to be resynchronized and
how far the synchronization has progressed. Both of these fields are updated as the
synchronization runs. Each time the a synchronization starts or is resumed, these fields will
be reset. In the case of a resume, the percent complete will reset to 0, but you should also
see a reduced total data out of synch.
To determine the megabytes of writes per second for each interval, run the performance tools
during a representative and peak period.
For example, if you determine that the amount of traffic is 5 MBps and you want to use
geographic mirroring for disaster recovery, then you need a pipe that can accommodate
5 MBps of data being transferred. If you are configuring two lines as data ports, then you
need 2.5 MBps per line.
From Table 8-7 on page 150, we can see the following facts:
A DS3/T3 allows 5.6 MBps theoretical throughput with a 2 MBps with a best practice at
30% utilization.
An OC-3 line allows 19.44 MBps theoretical throughput with 6 MBps with a best practice at
30% utilization.
You can initially start with two DS3 lines, but you might need to go to two OC-3 lines to
account and plan for growth.
To determine the time needed for initial synchronization, divide the total space utilized in the
IASP by the effective communications capability of the chosen communications lines. Speed
of the lines makes a big difference.
For example, if the IASP size is 900 GB and you are using 1 Gb Ethernet switches, then the
initial synchronization time will be less than an hour. However, if you are using two T3/DS3
lines, each having an effective throughput of 7.2 GB/hour, it would take around 63 hours to do
the initial synchronization. This was calculated by dividing the size of the IASP by the
effective GB/hour, that is, 900 GB divided by 14 GBps. A full resynchronization might also be
needed in the event of a disaster, so that must be factored into disaster recovery plans.
In most cases, the size of the data is used in the calculation, not the size of the IASP. An
exception to this is a *NWSSTG in an IASP. An *NWSSTG object is treated as one file, so the
size of the *NWSSTG is used instead of the amount of data within the *NWSSTG file.
To compute the initial synchronization time for *NWSSTG in an IASP, divide the size of the
network storage space of the IBM i hosted partition by the effective speed of the
communications mechanism.
For example, if the network storage space hosting IBM i was set up as 600 GB, it would take
42 hours to do the initial synchronization for a disaster recovery scenario using two DS3 lines.
Synchronization priority
The synchronization priority setting (low, medium, or high) determines the amount of
resources allocated to synchronization. Lower settings will gate synchronization, which will
also allow more resources to be allocated to non-synchronized work.
For additional information about best practices for high-availability environments, see
Chapter 15, “Best practices” on page 397.
The sizing methodology used is based on the observation that the processor time required to
perform I/O operating on the virtual SCSI server is fairly constant for a given I/O size.
Table 8-8 provides the approximate cycles per second for both physical disk and logical
volume operations on a 1.65 Ghz processor. For other frequencies, scaling by the ratio of the
frequencies is sufficiently accurate to produce a reasonable sizing.
For example, consider a Virtual I/O Server that uses two client logical partitions on physical
disk-backed storage. The first client logical partition requires a maximum of 7,000 8-KB
operations per second. The second client logical partition requires a maximum of 10,000
8-KB operations per second. The number of 1.65 Ghz processors for this requirement is
approximately ((7,000 × 47,000 + 10,000 × 47,000) / 1,650,000,000) = 0.48 processors,
which rounds up to a single processor when using a dedicated processor logical partition.
As a sizing general rule, a single dedicated POWER6 processor for VIOS is good enough for
about 40,000 virtual SCSI I/O.
For the memory sizing of VIOS, we do not need to consider the additional requirement for
VSCSI as similar to the processor because there is no data caching required by VSCSI. With
large I/O configurations and very high data rates, a 1 GB memory allocation for the VIOS is
likely to be sufficient. For low I/O rate situations with a small number of attached disks,
512 MB will most likely suffice.
The Virtual I/O Server requires a minimum of 30 GB of disk space with the Virtual I/O Server
Version 2.2.0.11, Fix Pack 24, Service Pack 1, or later.
On the VIOS host the mapped logical drives are seen as hdisks and then assigned to an
IBM i partition. We recommend that you assign entire hdisks to IBM i. VIOS supports logical
volumes where you can subset an hdisk into multiple volumes to assign to a partition. This is
not recommended for IBM i partitions.
IBM i has always been architected to perform best with more disk arms. This does not change
with SAN disks. You need to create a good number of logical drives, not one large drive.
As the LUNs are virtualized by VIOS, they do not have to match IBM i integrated disk sizes.
The technical minimum for any disk unit in IBM i is 160 MB and the maximum is 2 TB, as
measured in VIOS. Actual LUN size is based on the capacity and performance requirements
of each IBM i virtual client partition and load source disk restrictions (17.5 GB minimum,
1.9 TB maximum). IBM i performs best with logical drives that are the same size.
When creating an open storage LUN configuration for IBM i as a client of VIOS, it is crucial to
plan for both capacity and performance. As LUNs are virtualized for IBM i by VIOS instead of
being directly connected, it might seem that the virtualization layer will necessarily add a
significant performance overhead. However, internal IBM performance tests clearly show that
the VIOS layer adds a negligible amount of overhead to each I/O operation. Instead, the tests
demonstrate that when IBM i uses open storage LUNs virtualized by VIOS, performance is
almost entirely determined by the physical and logical configuration of the storage subsystem.
Consider that the bandwidth can handle peak write workload requirements.
For IBM Storage System, collect the performance data to get the highest write rate and
calculate the needed bandwidth as follows:
1. Assume 10 bits per byte for network overhead.
2. If the compression of devices for remote links is known, you can apply it.
3. Assume a maximum of 80% utilization of the network.
4. Apply a 10% uplift factor to the result to account for peaks in the 5-minute intervals of
collecting data, and a 20 - 25% uplift factor for 15-minute intervals.
As an example, we show how to calculate the required bandwidth for a given write workload:
1. The highest reported write rate at an IBM i is 40 MBps.
2. Assume 10 bits per byte for network overhead:
40 MBps * 1.25 = 50 MBps
3. Assume a maximum of 80% utilization of the network:
50 MBps * 1.25 = 62.5 MBps
4. Apply a 10% uplift for 5-minute intervals:
62.5 MBps * 1.1 = app 69 MBps
5. The needed bandwidth is 69 MBps.
A Recovery Point Object (RPO) estimation tool is available for IBM and IBM Business
Partners. This tool provides a method for estimating the RPO in a DS8000 Global Mirror
environment in relation to the bandwidth available and other environmental factors (see
Techdocs Document ID: PRS3246 for IBM and IBM Business Partners). For more
information, contact to IBM or IBM Business Partners.
FlashCopy SE is designed for temporary copies, so FlashCopy SE is optimized for use cases
where a small percentage of the source volume is updated during the life of the relationship. If
much more than 20% of the source is expected to change, there might be a trade-off in terms
of performance versus space efficiency. Also, the copy duration should generally not last
longer than 24 hours unless the source and target volumes have little write activity.
DS8000 FlashCopy SE
The DS8000 FlashCopy SE repository is an object within an extent pool and provides the
physical disk capacity that is reserved for space-efficient target volumes. When provisioning a
repository, storage pool striping will automatically be used with a multi-rank extent pool to
balance the load across the available disks. FlashCopy SE is optimized to work with
repository extent pools consisting of four RAID arrays. In general, we recommend that the
repository extent pool contain between one and eight RAID arrays. Extent pools larger than
eight RAID arrays are not recommended for the repository. It is also important that adequate
disk resources are configured to avoid creating a performance bottleneck. It is advisable to
use the same disk rotational speed or faster (10 K RPM or 15 K RPM) for the target repository
as for the source volumes. We also recommend that the repository extent pool have as many
disk drives as the source volumes.
After the repository is defined in the extent pool it cannot be expanded, so planning is
important to ensure that it is configured to be large enough. If the repository becomes full, the
FlashCopy SE relationships will fail. After the relationship fails, the target becomes
unavailable for reads or writes, but the source volume continues to be available for reads and
You can monitor and notify repository capacity and threshold using Simple Network
Management Protocol (SNMP) traps. You can set notification for any percentage of free
repository space with a default notification at 15% free and 0% free. Also, you can convert
and send these messages to QSYSOPR messages using the ACS toolkit. For more detailed
information, see in 10.2, “DS storage management” on page 178.
You can set the cache mode to readwrite for maximum performance when you create a
thin-provisioned volume. Also, to prevent a thin-provisioned volume from using up capacity
and getting offline, the autoexpand feature can be turned on.
In this chapter we introduce you to the new PowerHA GUI and discuss the other interfaces:
9.1, “Command line” on page 158
9.2, “PowerHA GUI” on page 160
9.3, “Cluster Resource Services GUI” on page 164
9.4, “High Availability Solution Manager GUI” on page 165
Note: The current release of IBM PowerHA for i GUI only supports the command-line
interface for working with SVC/V7000 copy services.
The IBM PowerHA for i licensed program provides IBM i command-line interfaces to
configure and manage your high-availability solution.
Selection or command
===>
When used with the *CREATE action, this command does as follows:
Creates the independent ASP using the specified non-configured disk units
Creates an ASP device description by the same name if one does not already exist
When used with the *DELETE action, this command does as follows:
Deletes the independent ASP
Deletes the ASP device description if it was created by this command
See 2.2, “Creating an IASP” on page 18, for more information about creating an
independent ASP.
Additional Parameters
Bottom
F3=Exit F4=Prompt F5=Refresh F12=Cancel F13=How to use this display
F24=More keys
Figure 9-2 Configure Device ASP (CFGDEVASP) command
Note: The Configure Device ASP (CFGDEVASP) command does not rely on SST/DST access.
However, the user profile must still have *IOSYSCFG and *SERVICE special authorities.
Additional Parameters
This new GUI makes configuration and management of a high availability environment easy
to use and intuitively obvious. The overall status of the high availability environment can be
easily determined and supports the ability to drill down to detect and correct specific problems
in the environment. The GUI enables basic functionality and allows configuration and
management of a geographic mirroring or DS8000-based replication environment.
Note: We recommend that PowerHA GUI is used for managing your high-availability
environment going forward, as the older Cluster Resource Services and High Availability
Solutions Manager GUIs will be discontinued after all of their functionality is brought into
the PowerHA GUI.
The PowerHA GUI is accessed via the IBM Systems Director Navigator for i from this website:
http://<system_name>:2001
You are then redirected automatically to port 2005, and your browser might issue a security
certificate warning. Choose Continue to this website and enter your user ID and password
to log in to IBM Systems Director Navigator for i. Expand the IBM i Management tasks tree to
see the new PowerHA option (Figure 9-4).
When PowerHA is accessed for the first time in the current session, it will look for any existing
configuration and inspect the HA environment (Figure 9-5).
See Chapter 11, “Creating a PowerHA base environment” on page 199, for the steps for
creating a PowerHA base environment.
Cluster properties
The cluster properties option of PowerHA GUI (Figure 9-6) gives you similar access as the
functions on the CHGCLU and CHGCLUVER command interfaces.
Note: The Current PowerHA version must be 2.1 for most of the options on PowerHA GUI
to work, including the Check Requirements function.
Note: Choosing the fix option from the PowerHA warning message panel modifies the
QGPL/QBATCH job queue entry for the QSYS/QBATCH subsystem from a shipped
default value of 1 to *NOMAX.
For more information about Cluster Resource Services GUI, see Chapter 7 of the IBM
Redbooks publication Implementing PowerHA for IBM i, SG24-7405:
http://www.redbooks.ibm.com/abstracts/sg247405.html
The High Availability Solutions Manager graphical interface (Figure 9-11 on page 166)
provides several predefined solutions. For each of these solutions, the dependent
technologies are configured based on your selection. The High Availability Solutions Manager
graphical interface provides easy-to-use tools to manage your high-availability solution.
Unlike the CRS GUI discussed in 9.3, “Cluster Resource Services GUI” on page 164, HASM
GUI employs a solution-based approach.The GUI operations are deployed via IBM Systems
Director Navigator for i, a web-based console that allows you to take the following actions:
Select a high-availability solution.
Verify requirements for your high-availability solutions.
Set up a high availability solution.
Manage a high availability solution.
For more information about configuring and managing your high-availability solution using the
HASM GUI, see Chapter 6 of Implementing PowerHA for IBM i, SG24-7405:
http://www.redbooks.ibm.com/abstracts/sg247405.html
In this chapter we describe the following major functional enhancements provided by ACS:
Easy-to-use interface for managing a DS storage and Copy Services environment
including externalized DSCLI commands, additional sanity checks, and scripting, which
can be used also for non-IBM i workloads (for example, WRKCSE and STRDSMGT)
Support for customized programming and faster resolution for Copy Services-related
issues by using default scripts provided and maintained for the Copy Services environment
Full automation of the FlashCopy process from the backup node
Additional safety measures and checks to prevent users from accessing inconsistent IASP
data after a FlashCopy relationship ended
Verification tests for PPRC switch-readiness, which can be user-initiated (CHKPPRC)
Ability to view and manage DS storage host connections and associated volumes
Simplified creation of new host connection when replacing a defective Fibre Channel IOA
Detailed audit trail for debugging issues (VIEWLOG)
Delivery of DS storage SNMP messages to the IBM i QSYSOPR message queue
Automated creation of a D volume FlashCopy at a Global Mirror target site
Support for Metro/Global Mirror (MGM) 3-site disaster recovery solutions
Support for integration with TPC-R in Metro Mirror or Metro/Global Mirror environments
ACS is implemented as a service offering from IBM STG Lab Services and is supported via
the regular IBM i software support structure. For further information about ACS see the IBM
STG Lab Services website for Power Systems here:
http://www-03.ibm.com/systems/services/labservices/platforms/labservices_power.html
ACS requires the user QLPAR on each IBM i cluster node and a user qlpar on the
DS6000/DS8000 systems. Instead of the user having to specify a user ID and password for
any of the ACS functions, it uses a DSCLI password file, which is set up only once from the
ACS storage management interface, as like described in 10.2, “DS storage management” on
page 178.
The main entry point to the ACS user interface for Copy Services is the Work with Copy
Services Environments accessed by WRKCSE (Figure 10-1).
DEMOGM GMIR
FLASH01 FLASH
FLASH02 FLASH
TOOLKIT FLASH
TOOLKIT MMIR
Bottom
Command
===>
F1=Help F3=Exit F4=Prompt F9=Viewlog F12=Cancel
Figure 10-1 ACS Copy Services Environments panel (WRKCSE)
As the available options from Work with Copy Services Environments shows, the user can
create a new Copy Services Environment or change, delete, display, or manage an existing
Copy Services Environment for Metro Mirror, Global Mirror, FlashCopy, or LUN-level switching.
Environment . . . . . . : TOOLKIT
Type . . . . . . . . . . : MMIR
ASP Device name . . . . : TOOLKIT
Source Copy Description : TKMMIRPS
Target Copy Description : TKMMIRPT
TPC Replication . . . . : *NO
Bottom
Volume relationships:
DEMOPROD DEMOHA
Volumes Volumes
7000-7002 7000-7002
7100-7102 7100-7102
Bottom
F1=Help F3=Exit F8=PPRC Paths F12=Cancel
Figure 10-2 ACS displaying a PPRC environment panel
The ACS has built-in checks to prevent the user from erroneously updating copy descriptions
by favoring the remote replication copy descriptions over FlashCopy descriptions and stores
all Copy Services environment data in the device domain data to make sure that it is kept
consistent within the IBM i cluster. It also provides protection against a user accidentally
manually changing the ASP copy descriptions instead of changing them via the ACS Copy
Services environment interface by refusing any ASP copy description changes with the
following message:
Copy Descriptions and ACS data do not match.
2. Pause
3. Resume
Selection
Using option 12 (Work with Volumes) for a Metro Mirror environment, the user can view the
PPRC state and suspend or resume Metro Mirror volume relationships (Figure 10-4).
Bottom
F1=Help F3=Exit F5=Refresh Status F9=Viewlog F12=Cancel
Figure 10-4 ACS Work with MMIR PPRC Volumes panel
However, while working within option 14 (List Stream files), ACS does not change the stream
files, so the user can still make adjustments before invoking a stream file as a DSCLI script
using option 9 (Run) with its output shown on the panel. This directory also has the *.result
files from any ACS command function, like CHKPPRC or SWPPRC, which is where a user
would find DSCLI errors if the command fails. The user can also run one of the existing
stream file profiles with option 9 (Run) to quickly get an interactive DSCLI session with the
corresponding DS storage system.
Bottom
F1=Help F3=Exit F5=Refresh Status F9=Viewlog F12=Cancel
Figure 10-6 ACS Available PPRC Paths panel
To easily set up Metro Mirror remote replication on the DS storage system with ACS, the user
can simply use a combination of the WRKCSE option 18 (Make PPRC Paths) and run the
corresponding mkpprc_from_PS.script that ACS does automatically generates based on the
DS storage system configuration information that the user has already provided when
creating the Copy Services environment.
For setting up Global Mirror remote replication on the DS storage system with ACS, the user
can simply invoke the mkpprc_GM_from_PS.script, the mksession scripts for both sides, the
chsession_GM_add_PS.script to add the primary volumes to the Global Mirror session, the
mkflash_GM_CG_PT.script to create the FlashCopy consistency group C volumes, and
finally the mkgmir_PS.script without ever needing to bother with manually entering complex
DSCLI commands.
Symmetric . . . . . . . : *YES
D-Copy Flash normal . . : *YES
D-Copy Flash reversed . : *NO
Extra CG Flash . . . . . : *NO
Override Master LSS . . : *YES
Source Mstr LSS . . . . : 02
More...
Volume relationships:
DEMOPROD DEMOHA DEMOHA DEMOPROD
PPRC Vols PPRC Vols CG Flash Vols CG Flash Vols
0200-0201 0200-0201 0250-0251 0250-0251
0300-0301 0300-0301 0350-0351 0350-0351
Bottom
F1=Help F3=Exit F8=PPRC Paths F12=Cancel
Figure 10-7 ACS Display a PPRC Environment panel for Global Mirror
2. Pause
3. Resume
4. Failover
5. Symmetrical switchover
Bottom
F3=Exit F4=Prompt F5=Refresh F10=Additional parameters F12=Cancel
F13=How to use this display F24=More keys
Figure 10-9 ACS Start a FlashCopy Backup panel (STRFLASH)
FlashCopy information:
FlashCopy node . . . . . . . . DEMOFC Name
Status . . . . . . . . . . . . *FLASHED *NONE, *FLASHED, number
Warm flash . . . . . . . . . . *YES *YES, *NO
Incremental flash . . . . . . *NO *YES, *NO
Quiesced flash . . . . . . . . *YES *YES, *NO
More...
F1=Help F3=Exit F12=Cancel
Figure 10-10 ACS Change CSE CRG Data panel (CHGCSEDTA)
At any time the user can verify the PPRC switch-readiness of an ACS Copy Services
environment by running CHKPPRC, which communicates with the DS storage system, similar to
displaying an ASP session in PowerHA, to verify that PPRC is running full-duplex and is set
up correctly with regard to the ACS PPRC setup.
Bottom
F3=Exit F4=Prompt F5=Refresh F12=Cancel F13=How to use this display
F24=More keys
Figure 10-11 ACS Switch PPRC command panel (SWPPRC)
When using the Auto replicate *DFT option, for a scheduled Metro Mirror switch, the
command looks at the CSE configuration parameter Automatic PPRC Replicate (whose
default value is *YES) to determine whether the PPRC direction will be established in the
reverse direction. For an unscheduled Metro Mirror switch, Auto replicate *DFT means *NO.
That is, no replication in the reverse direction is started. For Global Mirror it always changes
the replication direction for a scheduled switch, and also for an unscheduled switch if the
source DS storage system is available. ACS allows the user to also switch a paused PPRC
relationship, but only if all PPRC volumes are in the same paused state.
DMPINF allows the user to dump all ACS Copy Services environment information, such as
DSPCSEDTA and CSE scripts joblogs, to a stream file like
/QIBM/Qzrdhasm/qzrdhasm_DEMOPROD_110919_1202.txt.
Selection
Storage:
Device name . . . . . . . . IBM.2107-75AY031 Name
Primary HMC . . . . . . . . 9.5.168.55 IPv4
Second HMC . . . . . . . . . IPv4
SNMP Trap:
Message Queue Name . . . . . *SYSOPR name, *SYSOPR
Message Queue Library . . . name
Issue storage commands . . . *NO *YES, *NO
Selection
F3=Exit F12=Cancel
Figure 10-14 ACS Storage Authorization Management Menu
Bottom
Command
===>
F1=Help F3=Exit F5=Refresh F12=Cancel F17=Save F18=Display saved
F21=Print report
Figure 10-15 ACS Work with Disk Connections panel
Connection Details
Volume IDs . . : EA00 EA01 EA02 EA03 EB00 EB01 EB02 EB03
Bottom
F1=Help F3=Exit F12=Cancel
Figure 10-16 ACS Connection Details panel
Using F17 (Save) on the Work with Disk Connections panel allows the user to save a known
good configuration, so whenever a link failure occurs (represented by a **** not logged-in
information), the user can easily debug this connection from getting the original port login
information back with F18 (Display saved).
Option 11 (Replace IOA) on the Work with Disk Connections panel can be used to have the
host connections on the DS automatically updated (deleted and recreated) with the new
WWPN information after a replacement of a Fibre Channel IOA.
System i:
ASP list . . 179 ASP numbers
With the option F5 (Display report) a detailed log can be displayed (Example 10-1) where the
ACS expects IASP volumes in LSS 70/71 while the actual IASP 179 is configured on LSS 02/03.
Option 11 (Check System i Multipath) from the Storage Management Menu allows the user to
check whether the IASP disk units report in as multipath units since the last IPL or IASP
switch (that is, multi-path reset) (Example 10-2).
Bottom
Press Enter to continue.
Production Backup
LPAR A Volumes Global Copy B Volumes LPAR
FlashCopy FlashCopy
Consistency Consistency
Groups Groups
FlashCopy only FlashCopy
C Volumes C Volumes
after GM Failover LPAR
and FRR from B
D Volumes
SYSBAS
Configuration of a FashCopy D volume with ACS is done when creating a Global Mirror
environment (Figure 10-7 on page 173) by specifying D-Copy Flash normal *YES to set up a
FlashCopy D volume configuration on the Global Mirror target site, and by specifying D-Copy
Flash reversed *YES to allow for a FlashCopy on the target site also when the Global Mirror
direction is reversed (that is, the DS storage system at the target or secondary site acts as the
Global Mirror source).
Creating a FlashCopy to the D volumes, which involves an automatic prior Global Mirror
failover and FlashCopy fast-reverse-restore from the C volumes to the B volumes before
consistent data is on the B volumes, which can finally be flashed to the D volumes, can be
initiated by the user either with an option from the Global Mirror environment (Figure 10-8 on
page 174) or by using STRFLASH (Figure 10-9 on page 175).
IBM Power Storage TPC-R IBM Power Storage DS8000 IBM Power Storage
Primary Secondary Tertiary
Metro/Global Mirror Global Mirror
Session Session
MM Target/
MM Source Metro Mirror GC Source Global Copy GC Target
GM FLC CG
Figure 10-20 DS8000 3-site disaster recovery solution with Metro/Global Mirror
Other reasons for using TPC-R with ACS are if a customer likes to use Metro Mirror
consistency groups or uses TPC-R already with another environment and wants the IBM i
server platform included. Even with configured TPC-R support, for FlashCopy ACS always
uses its DSCLI interface. Similarly, ACS also does not make use of TPC-R’s Metro/Global
Mirror with Practice session, as it always fails over to the practice volume instead of the
GlobalCopy target volumes.
Environment . . . . . . : IASP01
TPC information:
TPC Replication . . . . *YES *YES, *NO
Metro-Global Mirroring *YES *YES, *NO
Primary server . . . . . 9.5.167.82 IPv4
Secondary server . . . . 9.5.167.57 IPv4
Session name . . . . . . MGM Name
User . . . . . . . . . . tpcruser Profile name
Password . . . . . . . .
More...
F1=Help F3=Exit F12=Cancel
Figure 10-22 ACS Change a GMIR Environment panel
ACS will then stop using the DSCLI and instead use its Java API interface into the TPC-R
server for managing the Copy Services functions. The CSE stream file scripts for DSCLI still
get created but without accompanying result files. Instead, there are Tpcr*Rtn.xml result files
showing the return status of the ACS Java API communication with the TPC-R server.
Note: Before planning to implement a Metro/Global Mirror environment with ACS for IBM i,
contact IBM STG Lab Services, as it requires careful planning to ensure that customer
requirements are met:
http://www-03.ibm.com/systems/services/labservices/contact.html
Using RUNDSCMD, a user can run a DSCLI script for retrieving the output in a comma-separated
text file that can be validated using a CL program, which in turn can make use again of the
existing ACS Copy Services environment stream files.
Bottom
F3=Exit F4=Prompt F5=Refresh F10=Additional parameters F12=Cancel
F13=How to use this display F24=More keys
Figure 10-23 ACS Run DS Scripted Command panel
Another example of using validation with RUNDSCMD is using its Summation column and CL
variable for returned total parameters to query the sum of certain values like the
out-of-sync tracks from all listed PPRC relationships (for example, to write a customized
program to predict the remaining synchronization time).
By using the Return column, Return key value, and CL variable for returned value in the
validation program, the user can search for the specified key value in the DSCLI output and
get the value at the specified return column returned in a CL variable for further processing in
customized programming.
Bottom
F3=Exit F4=Prompt F5=Refresh F12=Cancel F13=How to use this display
F24=More keys
Figure 10-24 ACS Run TPC Action command panel
RTVTPCCMD can be used to list the available TPC session actions as output in a CL variable,
which can be processed/verified in a CL program before running an available action to take
control of TPC-R from an IBM i client.
3
Partition returns to
1 IBM Power Systems production duties
Write "all" memory
content to disk
IBM System Storage
Production LPAR
2
Managing LPAR Sysbas
SYSBAS FlashCopy the disks
Flas
hCo
Backup LPAR
py
4 Sysbas’
SYSBAS
IPL the target
partition
5 6
Backup the The whole process is
target partition managed by the
to tape controller
FSFC is licensed separately from Advanced Copy Services for PowerHA but uses the same
Copy Services Environment menus for setting up the profiles and scripts to communicate with
the DS storage system. It needs to be installed on the managing and production partition in
the QZRDIASH5 library.
Note: While taking a warm FlashCopy with the full-system or IASP online, even when
quiesced, any object that is resident in memory and not journaled, such as files, data
queues, data areas, and so on, is subject to potential damage due to object
inconsistencies from parts of the object not on disk. Journaling allows the target partition to
recover from possible damaged journaled objects by applying journal entries to journaled
files during the IPL.
It is only with a full-system FlashCopy that there is a chance that on taking a warm
FlashCopy some IBM i internal objects can be found corrupted at IPL time. You might need
a SLIP install, or another FlashCopy might be needed for a recovery.
In addition to controlling the IBM i full-system FlashCopy process itself, FSFC also provides a
high level of integration with IBM i Backup Recovery and Media Services (BRMS) to make
sure that the backup by BRMS run from the FlashCopy backup partition appears in the BRMS
history containing all the backup and media information, as was done from the production
The FSFC full-system FlashCopy functions are configured on the managing partition using
CRTSYSCPY (Figure 10-27 on page 194). In our example we show a FlashCopy configuration
where CTCIHA7A is the production partition to be flashed via the managing partition to the
target partition CTCIHA7C with quiescing the production partition (OPTYPE *QUIESCE)
before taking the FlashCopy, resuming the production partition, and activating the backup
partition by starting its specified IP interface and BRMS backup job.
More...
Target partition host name . . . CTCIHA7C.RCHLAND.IBM.COM
More...
Target LPAR Device Config:
Backup device name . . . . . . TS3400 *NONE, device name
Backup device serial number . 78-1011003 *NONE, serial number
Robot host . . . . . . . . . . *NOCHANGE
Local internet address . . . . *NOCHANGE
+ for more values
Program to move QUSRBRM . . . . *SAVF_SOCK *SAVF_SOCK,
*VRTTAP,*NONE,name
Library . . . . . . . . . . . *LIBL *LIBL, library name
Save compression for QUSRBRM . . *DEV *DEV, *YES, *NO, *LOW...
Operation type . . . . . . . . . *QUIESCE *QUIESCE, *IPL, *NOIPL
Storage Type . . . . . . . . . . *DS8K *DS5K, *DS8K, *SVC
Issue IPL confirmation message *NO *YES, *NO
Source LPAR shutdown command . .
More...
F3=Exit F4=Prompt F5=Refresh F12=Cancel F13=How to use this display
F24=More keys
Figure 10-27 Full-system FlashCopy toolkit CRTSYSCPY command panel
More...
Target Comm Interfaces:
IO card serial number . . . . 00-EC7560A *NONE, Serial number
Line Description . . . . . . . FSFCL *NEW, line name
IO card port number . . . . . 0 0-32
IO card IP address . . . . . . 9.5.168.117
Network Mask . . . . . . . . . '255.255.255.0'
+ for more values
Target partition backup cmd . . STRBKUBRM USERDATA SBMJOB(*NO)
After the Copy Services environment has been created by using WRKCSE and the configuration
using CRTSYSCPY, a full-system FlashCopy can be started by using MKSYSCPY (Figure 10-29).
Bottom
F3=Exit F4=Prompt F5=Refresh F12=Cancel F13=How to use this display
F24=More keys
Figure 10-29 Full-system FlashCopy toolkit STRFLASH command panel
Part 3 Implementation
examples and best
practices
In this part, the final part of the book, we show you practical scenarios and
implementation examples with step-by-step instructions for how to configure and manage
your high-availability environment using IBM PowerHA SystemMirror for i.
We also discuss best practices to consider and follow in setting up and managing your
high-availability environment in Chapter 15, “Best practices” on page 397.
In this section, we set up a basic cluster environment with two nodes in it. We then create an
IASP on the production site. We set up advanced failure detection by adding a cluster monitor
and registering the cluster nodes with the HMC CIM server. In addition, we create an
administrative domain and add monitored resource entries to it. The entire setup is done
using the new PowerHA GUI. In addition, we provide you with the CL commands that you can
use alternatively to do this setup.
3. The wizard guides you through the steps necessary to create the cluster and asks you for
all information required for the setup. Figure 11-3 shows the steps.
Note: To have clustering on a node started automatically after an IPL, change the
system’s startup program with adding a STRCLUNOD entry.
Note: If you do not specify any failover message queue, then failover happens
immediately without any message being sent in advance.
9. You will see the page shown in Figure 11-9 after the cluster is created and all nodes are
started. The PowerHA GUI can then help you determine whether all requirements for
successfully running a cluster environment are met. Click Check Requirements to do so.
11.Clicking the toggle beside the message opens a Fix icon (Figure 11-11). If you click this
button corrective action is taken on the respective cluster node.
11.Provide the information shown in Figure 11-16. The CIM server host name is the IP name
of your HMC. The user ID and password are also those used to access the HMC.
The setup that we have done so far can also be done using the commands shown in
Example 11-1.
9. The IASP is created. The wizard regularly updates the completion status (Figure 11-28).
Bottom
F1=Help F9=Calculate Selection F11=View 2 F12=Cancel
Specifying *SELECT for the disk unit parameter shows the panel shown in Figure 11-30. It
provides you with a list of unconfigured disks on your system. Choose which ones you want to
add to your IASP and press Enter to create the IASP.
Resource
Opt Name Serial Number Type Model Capacity Rank Eligible
1 DD007 YQKJGD54BUK6 6B22 0050 19088 002 Yes
1 DD006 YDP4V2FVUK63 6B22 0050 19088 002 Yes
1 DD008 YUNHA7W9URJL 6B22 0050 19088 002 Yes
1 DD005 YWPZGH6N8LA9 6B22 0050 19088 002 Yes
Bottom
F1=Help F9=Calculate Selection F11=View 2 F12=Cancel
Configuration of ASP device IASP1 is 8% complete.
Figure 11-30 CFGDEVASP: Select disks to put into IASP
A message on the bottom of the page shows the progress of the IASP creation. Your age is
locked as long as the creation of the IASP is running.
2. To start the setup process choose Create Cluster Administrative Domain from the
pull-down menu (Figure 11-33).
4. After you have selected all nodes to be part of the administrative domain click OK
(Figure 11-35).
Figure 11-41 Cluster administrative domain: Add MRE for subsystem description
Figure 11-42 Cluster administrative domain: Add MRE for user profile
10.The monitored resource entries are added to the administrative domain and a first
synchronization occurs (Figure 11-43). If the MREs do not exist on any of the other nodes
in the administrative domain, they are automatically created.
The setup that we have done so far can also be done using the commands shown in
Example 11-2. Notice that you have to use individual commands for each monitored resource
that you want to add to the administrative domain.
Example 11-2 Command to create an administrative domain and add monitored resource entries
CRTCAD CLUSTER(PWRHA_CLU) ADMDMN(PWRHA_CAD) DMNNODL(CTCIHA9V CTCIHA9W)
SYNCOPT(*LASTCHG)
A geographic mirroring solution requires a cluster with a minimum of two cluster nodes, an
IASP, and optionally cluster monitors and an administrative domain. These components
make up the base environment for a geographic mirroring solution. The steps are discussed
in Chapter 11, “Creating a PowerHA base environment” on page 199.
Geographic mirroring is set up using the Make Highly Available wizard, which is found on
PowerHA GUI → Independent ASPs → Show All Others → pop menu of the iASP → Make
Highly Available (Figure 12-1).
3. The IASP must be part of a device cluster resource group (CRG) for setting up geographic
mirroring. On the Configuration page, you can either create a new device CRG or use an
existing one. For our scenario, we choose to create a new device CRG named
PWRHA_CRG. Click Next to work with the recovery domain.
5. Add backup node 1 and specify a name for the site and up to four data port IP addresses
(Figure 12-5).
7. If a device domain does not already exist, the wizard prompts you to and add the nodes to
a domain (Figure 12-7). Click OK to add them.
8. On the Devices panel, choose Modify from the IASP’s popup menu (Figure 12-9) to
specify a server takeover IP address and vary on options for the IASP.
10.You can specify an exit program to be executed after a switchover. This is optional, but if
specified, it must exist on all nodes in the recovery domain of the CRG. The exit program
can be used to set up an application environment and any other priming steps that are
necessary before users are allowed on the new production node after the switchover. For
our scenario, we leave the default Exit Program option as No and click Next
(Figure 12-11).
13.If there are no issues with the configuration, you will see a success dialog box
(Figure 12-14). PowerHA will add the nodes to the recovery domain, set up ASP copy
descriptions, and set up an ASP session for geographic mirroring.
The Configure Mirroring wizard can also be started using the IASP context menu
(Figure 12-16).
15.The Configure Mirroring wizard (Figure 12-18) can be used to set up an ASP session,
ASP copy descriptions, and so on, to be used for geographic mirroring. On the Welcome
page, click Next to choose the mirroring configuration options.
Figure 12-19 Specify mirroring type and a name for the ASP mirroring session
19.Specify names for the ASP copy descriptions and click Save (Figure 12-22).
Figure 12-24 Select disk units to create the IASP mirror copy
22.Verify the total disk capacity on the Selected Disk Units pane (Figure 12-25) and make
sure that you have adequate capacity to match the production copy. For more information
about sizing considerations refer to “Disk subsystem considerations” on page 146.
24.The Configure Mirroring wizard shows you a summary page with all the details for the
mirroring session, recovery domain, ASP copy descriptions, and the disk unit selection for
the mirror copy (Figure 12-28 on page 252).
In the following sections we show you how to manage various aspects of your high-availability
environment, including performing a planned switchover.
Monitored resources
A monitored resource is a system object or a set of attributes not associated with a specific
system object, such as the set of system environment variables. A resource represented by
an MRE is monitored for changes by a cluster administrative domain. When a monitored
resource is changed on one node in a cluster administrative domain, the change is
propagated to the other active nodes in the domain. You can access the Monitored
Resources panel from the Cluster Administrative Domain context menu (Figure 12-31).
On the Add Monitored Resources interface (Figure 12-33), you can type the name of the
resource or select from a list. When adding an MRE for a system object, the resource name is
the name of the system object. One or more attributes can be specified, and only attributes
that are specified will be monitored for changes.
On the Attributes interface, you can see which attributes are consistent and which are
inconsistent. As shown in Figure 12-35, you might find it easier to click the Global Status
column to sort the list and identify the attributes that are in an inconsistent state.
When you select the Switchover option, PowerHA gives you a preview of how the nodes and
their roles in a cluster resource group will look before and after a switchover (Figure 12-38).
Figure 12-38 Switchover summary with current and new node order
PowerHA shows you various progress messages and gives you a success dialog if
everything went well (Figure 12-39).
After you click Close on the Switch completed successfully dialog, you will be back on the
cluster resource groups interface. Click Refresh to view the updated role status for all the
nodes in the recovery domain (Figure 12-40).
To deconfigure geographic mirroring, the disk pool must be offline. First vary off the
independent ASP and then select Deconfigure Geographic Mirroring from the IASP pop-up
menu on the PowerHA GUI (Figure 12-42).
In various sections of this chapter, we discuss the following activities using an IBM i DS8000:
Setting up a Copy Services environment
Configuring Metro Mirror
Configuring FlashCopy
Configuring Global Mirror
Configuring LUN-level switching
Managing IBM i DS8000 Copy Services
The DS CLI installation package can be downloaded from the following URL:
ftp://ftp.software.ibm.com/storage/ds8000/updates/DS8K_Customer_Download_Files/CLI
The install packages are provided as ISO image files, which can be used by any virtual CD
drive tool on Windows.
For DS CLI installation procedures, refer to the IBM System Storage DS Command-Line
Interface User's Guide for the DS6000 series and DS8000 series, GC53-1127, included in the
install package and section 8.3 in IBM i and IBM System Storage: A Guide to Implementing
External Disks on IBM i, SG24-7120.
For the IBM i partitions to be able to manage DS8000 Copy Services, we need to provide a
DS storage user profile and its password. Whenever the password is changed on the
DS8000, the configuration on the IBM i partition needs to be changed at the same time. For
more information refer to section 9.5, “User management,” in IBM System Storage DS8000:
Architecture and Implementation, SG24-8886.
Note: User profiles and passwords are case sensitive on the DS8000.
The communication via DS CLI requires an IP connection between the IBM i partition and the
DS8000. This connection is initiated by IBM i to the DS8000, which listens on TCP port 1750.
Note: If there is a firewall between the IBM i partition and the DS8000, TCP port 1750 must
be authorized.
Environment overview
For this scenario, we need to create several cluster items. Cluster, cluster monitor, IASP, and
administrative domain setup are covered in Chapter 11, “Creating a PowerHA base
environment” on page 199. They are common to all scenarios that we describe in this chapter.
These are cluster items that we specifically configure for our Metro Mirror environment:
Cluster resource group
Device domain
ASPcopy descriptions and session
Takeover IP 10.0.0.1
Volume group ID V2 V2
Host connection ID 3 1
10.10.10.1
10.10.10.2
IBM Power
Systems server
DEMOPROD DEMOHA
IBM Power IBM i partition IBM i partition
Systems server
9.5.168.130
Production Site 9.5.168.129 Backup Site
Management network
9.5.168.55
9.5.168.55
IBM.2107-75AY031 IBM.2107-75AY032
60
01
60
02
60
00
60
01
60
02
61
01
61
02
61
00
61
01
61
02
A volumes B volumes
Note: Make sure that the paths are established as bi-directional, that is, on both DS8000s
from one to the other for the volumes’ LSS.
Note: For DS8000 releases prior to R4.3, to avoid a possible conflict with Volume Flash Init
DS8000 capabilities for volumes in a remote Copy Services relationship, the formatting of
volumes by DS8000 and IBM usually done when being added into an ASP must be
handled as follows:
1. Initialize and format disks using System Service Tools (STRSST):
a. Option 3. Work with disk units.
b. Option 3. Work with disk unit recovery.
c. Option 2. Disk unit recovery procedures.
d. Option 1. Initialize and format disk unit option. Notice that this operation is only a
marker, which disappears if there is an IPL.
2. Establish the Metro Mirror pair and wait for the source’s volumes to become synchronized.
3. Add volumes to the ASP or create the IASP to which volumes are added.
Figure 13-6 Specify type of mirroring and name of cluster resource group
Figure 13-7 Add the primary node to the cluster resource group
7. We continue by adding all necessary nodes to the cluster resource group. In our case
DEMOHA, which our backup node one, is located at the backup site with a site name of
SITE2 (Figure 13-8).
Figure 13-9 Continue when all nodes are in the cluster resource group
9. All nodes in a cluster resource group whose type is device must also be members of a
device domain. There is no starting point for creating a device domain in the PowerHA
GUI. Device domains are created by giving them a name (Figure 13-10). If there is already
an appropriate existing device domain, we can select it from the drop-down list. In our
environment, no device domain exists. Giving it a name and then clicking OK creates it.
11.When all nodes are included (Figure 13-12), we can click Close to continue.
Note: If PowerHA does not succeed to add all nodes into the device domain, you
receive a Cancel/Retry panel. This provides you with the opportunity to analyze the
reason for the failure, solve the cause, and retry adding nodes into the device domain.
Note: At this time, the cluster resource group is not yet created, but the device domain
has been created and nodes have been included.
Figure 13-13 All nodes are included in both a device domain and the cluster resource group
13.The next step is to launch the device configuration wizard. As shown in Figure 13-14,
select Modify for the device that you are working with (in our environment, IASP1).
The wizard displays several panels with the progress indicator (Figure 13-20).
20.The next panel asks whether you want to continue the wizard to create the ASP copy
descriptions for both independent ASP copies included in the cluster resource group. Click
Yes to launch the Configure Mirroring wizard (Figure 13-22).
24.The next step is related to creating ASP copy descriptions for both the source copy and
the target copy. As shown in Figure 13-27, select Modify Source Copy for the IASP (in
our case, this is still IASP1).
Note: For the DS8000 user ID and password, make sure that your browser settings do
not override the information that you provide with one that could already exist in the
saved passwords for Mozilla Firefox or AutoComplete settings for Microsoft Internet
Explorer, for example.
Figure 13-28 Specify first logical unit range for source copy description
26.You will need to add the logical unit ranges as needed. For each new range (in our case
6100-6102), click Add. When a range is not possible for a single volume, you still have to
enter it as a range, for example, 6000-6000.
Figure 13-29 Validate logical unit ranges for source copy description
29.When both source and target copy descriptions are complete (Figure 13-33), click Next to
review the configuration.
31.The wizard displays several panels with progress indicator (Figure 13-35).
Figure 13-35 Copy descriptions are being created and session started
Using CL commands
The corresponding CL commands for creating a device domain and a cluster resource group
are ADDDEVDMNE and CRTCRG.
The device domain must be properly populated with the nodes participating in the cluster
resource group before we can create it.
Example 13-3 Adding nodes to a device domain and creating ASP device description
ADDDEVDMNE CLUSTER(PWRHA_CLU) DEVDMN(PWRHA_DMN) NODE(DEMOPROD)
ADDDEVDMNE CLUSTER(PWRHA_CLU) DEVDMN(PWRHA_DMN) NODE(DEMOHA)
After the appropriate device domain and all IASP devices descriptions have been created, we
can create the cluster resource group (Example 13-4).
The related command for adding a Metro Mirror ASP copy description is ADDASPCPYD
(Example 13-5). You have to create one for each IASP copy (that is, one for the Metro Mirror
source and one for the target site).
Example 13-6 Starting the ASP session and cluster resource group
STRASPSSN SSN(IASP1_MM) TYPE(*METROMIR) ASPCPY((IASP1_MM1 IASP1_MM2))
STRCRG CLUSTER(PWRHA_CLU) CRG(PWRHA_CRG1)
Table 13-2 and Figure 13-38 on page 296 show the setup that we are going to use for our
FlashCopy environment.
Host connection ID 5 2
VolGrp: V26
WWPN 10000000C94122A2
DEMOHA A010 A012
A011 A013
FlashCopy
WWPN 10000000C9523E9D
DEMOFC A020
A022
A021
A023
VolGrp: V48
Bottom
F3=Exit F4=Prompt F5=Refresh F12=Cancel F13=How to use this display
F24=More keys
Figure 13-40 ADDASPCPYD for DEMOHA system
Bottom
F3=Exit F4=Prompt F5=Refresh F12=Cancel F13=How to use this display
F24=More keys
Bottom
F3=Exit F4=Prompt F5=Refresh F12=Cancel F13=How to use this display
F24=More keys
Figure 13-42 Start ASP session
Here we explain parameters used by STRASPSSN that might need more clarification:
– Session type
In our case we use *FLASHCOPY.
– Preferred source and target
This specifies the ASP copy descriptions the we created previously.The correct order
is important, as the source copy description points to the volumes that will be treated
as a source for the FlashCopy and the target description points to the target volumes.
Copy Descriptions
ASP
device Name Role State Node
IASP2 ASPCPYD1 SOURCE UNKNOWN DEMOHA
IASP2 ASPCPYD2 TARGET VARYOFF DEMOFC
Bottom
Press Enter to continue
Copy Descriptions
ASP
device Name Role State Node
IASP2 ASPCPYD1 SOURCE AVAILABLE DEMOHA
IASP2 ASPCPYD2 TARGET UNKNOWN DEMOFC
Bottom
Press Enter to continue
Figure 13-43 ASP session status for FlashCopy on target and source system
SITE 1 SITE 2
DEMOHA
DEMOPROD DEMOFC
IASP CPY2
Host connection ID 1 2
We have configured Metro Mirror replication between DEMOPROD and DEMOHA systems.
Figure 13-46 shows the status of this replication. On the DS storage system we can list the
volumes used by IASP1 on the DEMOHA system and their properties (we show only one
volume’s details) (Figure 13-47 on page 304).
Now we have to create the target volumes for the FlashCopy. The volumes need to be the
same type as the source ones, in this case A01. Using the DS CLI mkfbvol commands
(Figure 13-49), we create the volumes in the SE storage created earlier. We specify that the
volumes are track space efficient (TSE). After creating the FlashCopy target volumes we
assign them to a new volume group using the DS CLI mkvolgrp and chvolgrp command.
Figure 13-49 Creation of the volumes and adding them to a volume group
dscli> showhostconnect 2
Date/Time: October 5, 2011 6:40:18 PM CEST IBM DSCLI Version: 7.6.10.530 DS:
IBM.2107-75AY032
Name demofc_powerha
ID 0002
WWPN 10000000C9523E9D
HostType iSeries
LBS 520
addrDiscovery reportLUN
Profile IBM iSeries - OS/400
portgrp 0
volgrpID -
atchtopo -
ESSIOport all
speed Unknown
desc -
dscli> chhostconnect -volgrp v49 2
Date/Time: October 5, 2011 6:40:35 PM CEST IBM DSCLI Version: 7.6.10.530 DS:
IBM.2107-75AY032
CMUC00013I chhostconnect: Host connection 0002 successfully modified.
dscli> showhostconnect 2
Date/Time: October 5, 2011 6:40:39 PM CEST IBM DSCLI Version: 7.6.10.530 DS:
IBM.2107-75AY032
Name demofc_powerha
ID 0002
WWPN 10000000C9523E9D
HostType iSeries
LBS 520
addrDiscovery reportLUN
Profile IBM iSeries - OS/400
portgrp 0
volgrpID V49
atchtopo -
ESSIOport all
speed Unknown
desc -
Figure 13-50 Changing host connect for volume group V50 and DEMOFC system
Serial Part
Opt Description Type-Model Number Number
Combined Function IOP 2844-001 53-7141345 0000039J1719
Storage IOA 2787-001 1F-C5500BA 0000080P6417
Disk Unit 2107-A01 50-11A0BB8
Disk Unit 2107-A01 50-11A1BB8
Disk Unit 2107-A01 50-11A2BB8
Disk Unit 2107-A01 50-11A3BB8
Disk Unit 2107-A01 50-11A4BB8
Disk Unit 2107-A01 50-11A5BB8
Bottom
F3=Exit F4=Prompt F5=Refresh F12=Cancel F13=How to use this display
F24=More keys
Bottom
F3=Exit F4=Prompt F5=Refresh F12=Cancel F13=How to use this display
F24=More keys
Copy Descriptions
ASP
device Name Role State Node
IASP1 ASP1FC1 SOURCE UNKNOWN DEMOHA
IASP1 ASP1FC2 TARGET AVAILABLE DEMOFC
Bottom
Press Enter to continue
Table 13-4 and Figure 13-56 on page 313 describe and show our environment.
Takeover IP 10.0.0.1
Host connection ID 07 01
10.10.10.1
10.10.10.2
IBM Power
Systems server
DEMOPROD DEMOHA
IBM Power IBM i partition IBM i partition
Systems server
9.5.168.130
Production Site Backup Site
Management network
9.5.168.129
9.5.168.129
IBM.2107-13ABGAA IBM.2107-1375210
02
01
02
00
02
01
03
01
03
00
03
01
Flashcopy in
A volumes B volumes
normal
02
50
02
51
02
50
02
51
03
50
03
51
03
50
03
51
D volumes C volumes
Note: Make sure that the paths exist on both DS8000s from one to the other for the
volumes’ LSS.
Note: In a Global Copy relationship, the source volumes remain at copy pending status
and the target volumes at target copy pending status. They never reach full duplex and
target full duplex, because of the asynchronous process.
Using CL commands
The corresponding CL commands for creating a device domain and a cluster resource group
for a Global Mirror scenario are exactly the same as the ones used for the Metro Mirror
scenario. Refer to “Using CL commands” on page 293 for more information.
The related command for adding a Global Mirror ASP copy description is ADDASPCPYD
(Example 13-11). You have to create one for each IASP copy (that is, one for the Global Mirror
source and one for the target site). The difference with the Metro Mirror ASP copy description
creation command resides in the consistency group volume specification. In our case,
volumes 0250 - 0251 and 0350 - 0351 are those volumes on both source and target Global
Mirror copy descriptions.
Note: STRASPSSN fails if at least one of the consistency group volumes is included in a
volume group. PowerHA assumes that there is a DS8000 configuration error in this case
because the consistency group volumes must not be accessed by any host, which is
supposed to be the case if they are members of a volume group.
Example 13-12 Starting the ASP session and cluster resource group
STRASPSSN SSN(IASP2_GM) TYPE(*GLOBALMIR) ASPCPY((IASP2_GM1 IASP2_GM2))
STRCRG CLUSTER(PWRHA_CLU) CRG(PWRHA_CRG2)
Cluster items that we specifically configure for our LUN-level switching environment are
cluster resource groups.
Environment overview
In this scenario, we use and switch DS8000 LUNs with volume group ID V48 between
partition DEMOFC and partition DEMOHA. Table 13-5 provides more detailed information
about the configuration.
IASP IASP_LUN
DS8000 IP 9.5.168.55
IASP_LUN
Backup Site
WWPN : 10000000C94122A2
DEMOHA
*SYSBAS
Note: When configuring LUN-level switching, a volume group is assigned only to the IBM i
production node. The host connection for the backup node is not allowed to have a volume
group assigned, as it will be automatically assigned when LUNs are switched by PowerHA.
Example 13-13 Information about volume groups and host connection on DS8000
dscli> showvolgrp V48
Date/Time: September 29, 2011 9:28:20 AM CDT IBM DSCLI Version: 7.6.10.530 DS:
IBM.2107-75AY032
Name V48fc
ID V48
Type OS400 Mask
Vols A020 A021 A022 A023
dscli> lshostconnect 2 5
Date/Time: September 28, 2011 11:27:31 AM CDT IBM DSCLI Version: 7.6.10.530 DS:
IBM.2107-75AY032
Name ID WWPN HostType Profile portgrp volgrpID
ESSIOport
Configuration on IBM i
Create a basic cluster environment for LUN-level switching, as described in Chapter 11,
“Creating a PowerHA base environment” on page 199:
1. Create a cluster with two nodes, DEMOFC and DEMOHA, and a device domain
(Example 13-14). The cluster nodes are automatically started as the default setting.
2. For production node DEMOFC, create an IASP using LUNs previously assigned by
DS8000. For backup node DEMOHA we create an IASP device description with the same
name that we used on the production node (Example 13-15).
On Backup site
CRTDEVASP DEVD(IASP_LUN) RSRCNAME(IASP_LUN)
3. After creating a cluster environment and IASP, we create a device cluster resource group
using CRTCRG (Example 13-16). At this time do not list the IASP as a configuration object
for the CRG.
Note: Make sure that you set the same site name for each recovery domain node. In
our scenario we use the site name SITE1 for node DEMOFC and node DEMOHA.
On this command, specify the host identifiers and volume groups for each node in the
recovery domain parameter (RCYDMN), even though we do not make the volume group
assign a host connection ID for the backup node DEMOHA (Example 13-13 on page 318).
5. Add the IASP to the configuration object list of device CRGs created in step 3 using
ADDCRGDEVE (Example 13-18). The reason why we did not add the configuration object list
when we created the CRG is that we migth get misleading error messages in the job log.
This happens when the IASP is listed in the configuration object list of CRGs before the
copy description is created. After the copy description is created, PowerHA tells the CRG
that PowerHA is in control of the switching, which causes the CRG to skip its IOP and
tower switching checks. If the IASP is listed in the configuration object list of CRGs before
the copy description is created, then we see misleading error messages in the job log.
After the copy description is created, PowerHA tells the CRG that PowerHA is in control of
the switching, which causes the CRG to skip its IOP and tower switching checks.
6. After adding the IASP to the configuration object list of the device CRGs, we start the CRG
for LUN-level switching between the production site and the backup site using STRCRG
(Example 13-19).
Now your IBM PowerHA SystemMirror for i DS8000 LUN-level switching configuration is
complete and your IASP is highly available for planned and unplanned switches, as described
in 13.2.2, “Using CL commands for DS8000 LUN-level switching” on page 346.
13.2.1 Switchover and switchback for a Metro Mirror or Global Mirror planned
outage
In this section we describe the procedure for a planned site switch for Metro Mirror using the
PowerHA GUI, and for Global Mirror using CL commands.
From the user standpoint, the steps for performing a switch for both scenarios are the same.
The difference is only related to DS8000 commands done under-the-covers by PowerHA to
perform the actions.
Note: Before starting any switchover make sure that no jobs are using the IASP any longer.
Before starting the switch, we look at our current configuration. DEMOPROD is the current
production system, and DEMOHA is the current first backup system.
In Figure 13-58 and Figure 13-59 on page 322, you can see the IP address configuration on
both DEMOPROD and DEMOHA systems. On DEMOPROD the takeover IP address 10.0.0.1
is active. On DEMOHA, this IP address is inactive. The takeover IP address is usually the one
to which all users and systems connect.
Bottom
F3=Exit F5=Refresh F6=Print list F11=Display line information
F12=Cancel F17=Top F18=Bottom
Figure 13-58 Metro Mirror: IP configuration on preferred production before switchover
Bottom
F3=Exit F5=Refresh F6=Print list F11=Display line information
F12=Cancel F17=Top F18=Bottom
Figure 13-59 Metro Mirror: IP configuration on preferred backup before switchover
Figure 13-60 shows the current status for DS volume members of the Metro Mirror
relationships. With lspprc on the current production DS8000 75AY031, all volumes are full
duplex, which is the normal status for source volumes. On the current backup DS8000
75AY032, all volumes are target full duplex, which is the normal status for target volumes.
Note: Any cluster node can be selected to perform a switchover as long as it is in active status.
1. Switchover activity is related to the cluster resource group. Therefore, to enter the
switchover wizard (Figure 13-61), click Cluster Resource Groups.
3. Before starting the switchover, we have the opportunity to review the CRG and recovery
domain current status, and what will be the result of the operation. In our case
(Figure 13-63), all statuses are correct, and after the switchover DEMOPROD will be the
first backup site and DEMOHA will be the production site. If everything is correct, click OK.
4. The wizard displays the switchover progress. There are seven steps (Figure 13-65):
a. The first step ends all jobs using the IASP.
b. The second step varies off the IASP on the production system.
Step Elapsed
time
Cluster vary job submission
Ending jobs using the ASP 00:00:00
Waiting for jobs to end 00:00:04
Image catalog synchronization
> Writing changes to disk 00:00:08
Step Elapsed
time
UID/GID mismatch correction 00:00:00
Database access path recovery 00:00:01
> Database cross-reference file merge 00:00:01
SPOOL initialization
Image catalog synchronization
Command analyzer recovery
Catalog validation
g. After the seventh step has completed, which is starting the takeover IP address on the
backup system, the overall switchover is complete (Figure 13-69).
Figure 13-70 Cluster Resource Group panel after completed Metro Mirror switchover
6. We now have the correct status for the cluster resource group that we just switched over
from production to backup, from the DEMOPROD to the DEMOHA node (Figure 13-71).
Figure 13-71 Metro Mirror cluster resource group status after switchover
After the switchover is complete, we can compare the situation to the one before the
switchover. DEMOPROD is the current first backup system, and DEMOHA the current
production system.
Bottom
F3=Exit F5=Refresh F6=Print list F11=Display line information
F12=Cancel F17=Top F18=Bottom
Figure 13-72 Metro Mirror: IP configuration on preferred production after switchover
Bottom
F3=Exit F5=Refresh F6=Print list F11=Display line information
F12=Cancel F17=Top F18=Bottom
Figure 13-73 Metro Mirror: IP configuration on preferred backup after switchover
Figure 13-74 shows the current situation for DS volume members of the Metro Mirror
relationships. With lspprc on the current backup DS8000 75AY031, all volumes are target full
duplex, which is the normal status for target volumes. On the current production DS8000
75AY032, all volumes are full duplex, which is the normal status for source volumes.
Note: In contrast to the GUI, the CL command will not give you any possibility to switch
over or switch back if there is one incorrect item.
Before activating the switch, we can check whether everything is fine from a clustering state
perspective and that nothing will prevent the switchover from succeeding:
1. The first items that we check are the cluster consistency and the cluster resource group
status. Use WRKCLU, then option 9 (Work with cluster resource groups) (Figure 13-75).
Selection or command
===> 9
Cluster Primary
Opt Resource Group Type Status Node
Bottom
Parameters for options 1, 2, 3, 8, 9 and 20 or command
===>
F1=Help F3=Exit F4=Prompt F5=Refresh F9=Retrieve F12=Cancel
F13=Work with cluster menu
Figure 13-76 Global Mirror switch over Cluster Resource Group panel
Bottom
Parameters for options 1 and 20 or command
===>
F1=Help F3=Exit F4=Prompt F5=Refresh F9=Retrieve F12=Cancel
F13=Work with cluster menu
Cluster . . . . . . . . . . . . : PWRHA_CLU
Cluster resource group . . . . . : PWRHA_CRG2
Reporting node . . . . . . . . . : DEMOHA
Consistent information in cluster: Yes
More...
Cluster . . . . . . . . . . . . : PWRHA_CLU
Cluster resource group . . . . . : PWRHA_CRG2
Reporting node . . . . . . . . . : DEMOHA
Consistent information in cluster: Yes
Figure 13-78 Global Mirror switchover DSPCRGINF command first and second panel
Cluster . . . . . . . . . . . . : PWRHA_CLU
Cluster resource group . . . . . : PWRHA_CRG2
Reporting node . . . . . . . . . : DEMOHA
Consistent information in cluster: Yes
Bottom
Number of recovery domain nodes : 2
Cluster Primary
Opt Resource Group Type Status Node
Bottom
Parameters for options 1, 2, 3, 8, 9 and 20 or command
===>
F1=Help F3=Exit F4=Prompt F5=Refresh F9=Retrieve F12=Cancel
F13=Work with cluster menu
Figure 13-80 Global Mirror switch over from Cluster Resource Group panel
Bottom
Parameters for options 1 and 20 or command
===>
F1=Help F3=Exit F4=Prompt F5=Refresh F9=Retrieve F12=Cancel
F13=Work with cluster menu
During the switchover, on both DS8000s the following actions are submitted by PowerHA
during step 4 (Changing Recovery Domain) reported by the GUI:
a. Check the Global Copy status on the preferred source DS8000.
b. Check the Global Mirror LSS status on the preferred source DS8000.
c. Pause the Global Mirror session on the preferred source DS8000.
d. Remove volumes from the Global Mirror session on the preferred source DS8000.
e. Resume the Global Mirror session on the preferred source DS8000.
f. Fail over the Global Copy relationships on the preferred target DS8000.
g. Fast reverse restore FlashCopy relationships on the preferred target DS8000 to put
consistent data on newly Global Copy source volumes.
The persistent option is not set in the DS8000 command, which means that the
FlashCopy relationships no longer exist at the end of reverseflash.
h. Fail back Global Copy relationships on the preferred target DS8000 to synchronize
the current production volumes with the current backup volumes on the preferred
source DS8000.
i. Create the FlashCopy relationships on the current target DS8000 to build a
consistency group.
j. Update the Global Mirror session on the current source DS8000 to include the current
source volumes.
k. Start the Global Mirror session on the current source DS8000.
The main difference is that D volumes do not exist when compared to our scenario setup in
Figure 13-56 on page 313. Therefore, when creating the ASP copy description for the
preferred source IASP, we do not specify the consistency group volumes (Example 13-20).
Example 13-20 Source ASP copy description creation command for an asymmetrical Global Mirror
ADDASPCPYD ASPCPY(IASP2_GM1) ASPDEV(IASP2) CRG(PWRHA_CRG2) LOCATION(*DEFAULT)
SITE(SITE1) STGHOST('qlpar' 'password' ('9.5.168.32'))
LUN('IBM.2107-13ABGAA' ('0200-0201' '0300-0301') ())
For a switchover to the preferred backup site in an asymmetrical Global Mirror configuration,
PowerHA cannot reverse Global Mirror because there have been no FlashCopy consistency
group volumes configured on the preferred source DS8000. Therefore, at the end of the
switchover the Global Copy relationship volume state on the preferred backup DS8000,
which is the current production DS8000, is source suspended. They are not members of the
Global Mirror session on the preferred backup DS8000 and the FlashCopy relationships no
longer exist.
Bottom
Parameters for options 1 and 20 or command
===>
F1=Help F3=Exit F4=Prompt F5=Refresh F9=Retrieve F12=Cancel
F13=Work with cluster menu
The message suggests that you either perform a resume or a reattach operation on the ASP.
In our case, the mirroring is not only suspended, but it is also in a failover status. This means
that we need to run a reattach operation for the ASP session. However, we also need to
establish the Global Mirror session in the proper direction again. This is the reason why in this
special case, we have to run the following operations:
1. Make sure that independent ASPs are varied off on both sides.
2. On the current production DS, perform failbackpprc. This synchronizes data from the
current production DS to the current backup DS (Example 13-21).
Example 13-21 Asymmetrical Global Mirror switch back, synchronize preferred DS with current DS
dscli> failbackpprc -dev IBM.2107-1375210 -remotedev IBM.2107-13ABGAA -type gcp 0200-0201:0200-0201
0300-0301:0300-0301
Date/Time: October 4, 2011 3:48:19 PM CDT IBM DSCLI Version: 7.6.10.530 DS: IBM.2107-1375210
CMUC00197I failbackpprc: Remote Mirror and Copy pair 0200:0200 successfully failed back.
CMUC00197I failbackpprc: Remote Mirror and Copy pair 0201:0201 successfully failed back.
CMUC00197I failbackpprc: Remote Mirror and Copy pair 0300:0300 successfully failed back.
CMUC00197I failbackpprc: Remote Mirror and Copy pair 0301:0301 successfully failed back.
dscli> lspprc 0200-03FF
Date/Time: October 4, 2011 3:48:23 PM CDT IBM DSCLI Version: 7.6.10.530 DS: IBM.2107-1375210
ID State Reason Type SourceLSS Timeout (secs) Critical Mode First Pass Status
==================================================================================================
0200:0200 Copy Pending - Global Copy 02 60 Disabled True
0201:0201 Copy Pending - Global Copy 02 60 Disabled True
3. On the current production node make sure that the replication is finished, as indicated by a
synchronization progress of 100 in the DSPASPSSN command panel (Figure 13-86).
Copy Descriptions
ASP
device Name Role State Node
IASP2 IASP2_GM2 SOURCE VARYOFF DEMOHA
IASP2 IASP2_GM1 TARGET RESUMING DEMOPROD
Bottom
Press Enter to continue
Figure 13-86 Asymmetrical Global Mirror switchback, synchronizing back current to preferred
production DS8000
4. Make the preferred source DS8000 the source of the Global Copy replication. On the
preferred source DS8000, run the pair of failoverpprc/failbackpprc commands
(Example 13-22).
Example 13-22 Asymmetrical Global Mirror switchback, making the preferred source DS8000 the current source
dscli> lspprc 0200-03FF
Date/Time: October 4, 2011 4:02:03 PM CDT IBM DSCLI Version: 7.6.10.530 DS: IBM.2107-13ABGAA
ID State Reason Type SourceLSS Timeout (secs) Critical Mode First Pass Status
=========================================================================================================
0200:0200 Target Copy Pending - Global Copy 02 unknown Disabled Invalid
0201:0201 Target Copy Pending - Global Copy 02 unknown Disabled Invalid
0300:0300 Target Copy Pending - Global Copy 03 unknown Disabled Invalid
0301:0301 Target Copy Pending - Global Copy 03 unknown Disabled Invalid
dscli> failoverpprc -dev IBM.2107-13ABGAA -remotedev IBM.2107-1375210 -type gcp 0200-0201:0200-0201
0300-0301:0300-0301
Date/Time: October 4, 2011 4:09:18 PM CDT IBM DSCLI Version: 7.6.10.530 DS: IBM.2107-13ABGAA
CMUC00196I failoverpprc: Remote Mirror and Copy pair 0200:0200 successfully reversed.
CMUC00196I failoverpprc: Remote Mirror and Copy pair 0201:0201 successfully reversed.
CMUC00196I failoverpprc: Remote Mirror and Copy pair 0300:0300 successfully reversed.
CMUC00196I failoverpprc: Remote Mirror and Copy pair 0301:0301 successfully reversed.
dscli> failbackpprc -dev IBM.2107-13ABGAA -remotedev IBM.2107-1375210 -type gcp 0200-0201:0200-0201
0300-0301:0300-0301
Date/Time: October 4, 2011 4:09:30 PM CDT IBM DSCLI Version: 7.6.10.530 DS: IBM.2107-13ABGAA
CMUC00197I failbackpprc: Remote Mirror and Copy pair 0200:0200 successfully failed back.
5. The Global Mirror configuration also needs to be built again on the preferred source
DS8000. Refer to the following sections:
– “Making a Global Mirror session” on page 315
The sessions should already exist. Just replace the mksession commands with
chsession ones.
– “Creating the FlashCopy relationships on the backup site” on page 314
– “Starting Global Mirror session” on page 316.
Note: You might need to remove the Global Mirror session using the rmgmir DS8000
command before being allowed to create it again.
Example 13-23 Asymmetrical Global Mirror failback: Change CRG role command
CHGCRG CLUSTER(PWRHA_CLU) CRG(PWRHA_CRG2) CRGTYPE(*DEV) RCYDMNACN(*CHGCUR)
RCYDMN( (DEMOPROD *PRIMARY *SAME SITE1 *SAME *SAME)
(DEMOHA *BACKUP 1 SITE2 *SAME *SAME))
After completing these steps you have re-established Global Mirror in its original direction
from the preferred source to the preferred target DS8000.
Note: Before starting any switchover make sure that no jobs are using the IASP anymore.
1. Make sure that “Consistent information in cluster” is Yes. Set the CRG status (in our case
PWRHA_CRG3) to Active (Figure 13-87) by using WRKCLU, then option 9 (Work with cluster
resource groups).
Cluster Primary
Opt Resource Group Type Status Node
Bottom
Parameters for options 1, 2, 3, 8, 9 and 20 or command
===>
F1=Help F3=Exit F4=Prompt F5=Refresh F9=Retrieve F12=Cancel
F13=Work with cluster menu
Bottom
Parameters for options 1 and 20 or command
===>
F1=Help F3=Exit F4=Prompt F5=Refresh F9=Retrieve F12=Cancel
F13=Work with cluster menu
Alternatively, use DSPCRGINF to summarize the information and status of the CRG.
3. Use CHGCRGPRI to switch the IASP from the primary node to the backup node
(Example 13-24).
Cluster Primary
Opt Resource Group Type Status Node
Bottom
Parameters for options 1, 2, 3, 8, 9 and 20 or command
===>
F1=Help F3=Exit F4=Prompt F5=Refresh F9=Retrieve F12=Cancel
F13=Work with cluster menu
Cluster Resource Services API QcstInitiateSwitchOver completed. +
You can verify that the Recovery Domain nodes are in the expected active status and that
their role is the preferred backup status. In our case, DEMOHA is the production node and
DEMOFC is the backup node (Figure 13-91).
Bottom
Parameters for options 1 and 20 or command
===>
F1=Help F3=Exit F4=Prompt F5=Refresh F9=Retrieve F12=Cancel
F13=Work with cluster menu
5. Verify whether the IASP has been varied on (that is, that it is in available status) at the
backup node using the following command:
WRKCFGSTS CFGTYPE(*DEV) CFGD(*ASP)
If the configuration object online parameter in the device CRG is set to *ONLINE,
PowerHA automatically varies on the IASP on node at a switch over.
Display Messages
System: DEMOHA
Queue . . . . . : QSYSOPR Program . . . . : *DSPMSG
Library . . . : QSYS Library . . . :
Severity . . . : 99 Delivery . . . : *HOLD
Bottom
F3=Exit F11=Remove a message F12=Cancel
F13=Remove all F16=Remove all except unanswered F24=More keys
The IASP (in our case IASP1) is varied on for the node at the secondary site in our
example, as we used the CRG configuration object *ONLINE setting. Also, the takeover IP
address is started on the secondary site.
Cluster node roles are changed so that the node at the secondary site becomes the
primary and the node at the failed primary site becomes the backup (Figure 13-94), which
is the recovery domain for our cluster resource group PWRHA_CRG1. In our case,
DEMOHA becomes the primary node and DEMOPROD becomes the first backup node
with an inactive status.
Bottom
Parameters for options 1 and 20 or command
===>
F1=Help F3=Exit F4=Prompt F5=Refresh F9=Retrieve F12=Cancel
F13=Work with cluster menu
Bottom
Parameters for options 1 and 20 or command
===>
F1=Help F3=Exit F4=Prompt F5=Refresh F9=Retrieve F12=Cancel
F13=Work with cluster menu
Copy Descriptions
ASP
device Name Role State Node
IASP1 IASP1_MM2 SOURCE UNKNOWN DEMOHA
IASP1 IASP1_MM1 DETACHED SUSPENDED DEMOPROD
Bottom
Press Enter to continue
d. Establish back the replication from the preferred backup to the preferred production DS.
i. For Metro Mirror or symmetrical Global Mirror, re-attach the detached ASP session
on the preferred primary node. That is, start the Metro Mirror (or symmetrical Global
Mirror) replication from the preferred secondary DS8000 to the preferred primary
DS8000, with the following command:
CHGASPSSN SSN(IASP1_MM) OPTION(*REATTACH) ASPCPY((IASP1_MM1 IASP1_MM2))
By default, this command requires an answer to the following message in the
QSYSOPR message queue:
HAA2000 “Reattach of ASP session IASP1_MM was requested. (C G)”
Bottom
Type reply below, then press Enter.
Reply . . . .
ii. For asymmetrical Global Mirror, start the replication from the preferred secondary
DS8000 to the preferred primary DS8000. This must be done with failbackpprc on
the preferred secondary DS8000 (Example 13-21 on page 343).
Copy Descriptions
ASP
device Name Role State Node
IASP1 IASP1_MM2 SOURCE AVAILABLE DEMOHA
IASP1 IASP1_MM1 TARGET ACTIVE DEMOPROD
Bottom
Press Enter to continue
Copy Descriptions
ASP
device Name Role State Node
IASP2 IASP2_GM2 SOURCE AVAILABLE DEMOHA
IASP2 IASP2_GM1 TARGET RESUMING DEMOPROD
Bottom
Press Enter to continue
Now we can perform a switchback to the preferred primary node, as described in section
13.2.1, “Switchover and switchback for a Metro Mirror or Global Mirror planned outage” on
page 320.
To simulate a primary node failure without a panic message, we force an immediate
power off from the HMC on primary node partition. This action simulates a server failure.
In this case, the primary node is not able to send a panic message to the backup node.
Bottom
Parameters for options 1, 2, 9 and 20 or command
===>
F1=Help F3=Exit F4=Prompt F5=Refresh F9=Retrieve
F11=Order by status F12=Cancel F13=Work with cluster menu
After bringing back the preferred production node, he steps to fail back are the same as in
our first case above.
If you are not using the advanced node failure detection feature, the backup node is
unable to know the real production node status, and it puts the production node in the
partition status within the recovery domain. We describe this behavior in the third case,
below.
To simulate a cluster partition condition, we break the heartbeat IP connection by ending
the heartbeat IP interface on the primary node.
If the issue is either a real production node issue, or, for example, a network issue that
prevents any user from connecting to the production node, you might decide to perform
a failover.
The node status for the primary cluster node needs to be changed from the partition to
failed with the following command:
CHGCLUNODE CLUSTER(PWRHA_CLU) NODE(DEMOPROD) OPTION(*CHGSTS)
Bottom
Parameters for options 1 and 20 or command
===>
F1=Help F3=Exit F4=Prompt F5=Refresh F9=Retrieve F12=Cancel
F13=Work with cluster menu
The remaining actions, such as varying on the independent ASP and starting the takeover
IP address, must be done manually.
After recovery of the preferred primary node, the steps to fail back are the same as for our
first case above.
DEMOFC Active
DEMOHA Active PWRHA_DMN
DEMOPROD Partition PWRHA_DMN
Bottom
Parameters for options 1, 2, 9 and 20 or command
===>
F1=Help F3=Exit F4=Prompt F5=Refresh F9=Retrieve
F11=Order by status F12=Cancel F13=Work with cluster menu
Bottom
Parameters for options 1 and 20 or command
===>
F1=Help F3=Exit F4=Prompt F5=Refresh F9=Retrieve F12=Cancel
F13=Work with cluster menu
Message . . . . : Automatic fail over not started for cluster resource group
PWRHA_CRG1 in cluster PWRHA_CLU.
Cause . . . . . : A failure has occurred on primary node DEMOPROD or with
the job associated with cluster resource group PWRHA_CRG1 on node DEMOPROD.
Cluster Resource Services is unable to determine the state of the resources
being managed by cluster resource group PWRHA_CRG1 and is unable to perform
automatic fail over. The type of failure is reason code 1. Possible reason
codes are:
1 -- The cluster has become partitioned because the failure appears to be
a communication failure.
2 -- Either there is no backup node defined in the cluster resource group
or all backup nodes are not active. If this is a resilient device cluster
More...
Press Enter to continue.
If the issue is really only about heartbeat communication and does not impact production
activity, and if it can be fixed quickly, after it is fixed, after a 15-minute interval (the default
value), the cluster status will automatically change back to active.
Note: To get a consistent status of your application data with DS8000 Remote Copy
Services, the IASP needs to be varied off before being detached.
Caution: While the ASP session is detached, it is not possible to perform a switch.
Before starting the operations, make sure that everything is correct for the cluster, cluster
resource groups, and recovery domain, as described in previous sections:
1. Vary off the independent ASP (in our case IASP1) on the production node using the
following command:
VRYCFG CFGOBJ(IASP1) CFGTYPE(*DEV) STATUS(*OFF)
2. Detach the ASP session, using the following command:
CHGASPSSN SSN(IASP1_MM) OPTION(*DETACH)
Note: The DETACH option is required to run on the backup node for a Global Mirror
configuration and on the production node for a Metro Mirror configuration.
3. Vary on the independent ASP (in our case IASP1) on the production node using the
following command:
VRYCFG CFGOBJ(IASP1) CFGTYPE(*DEV) STATUS(*ON)
Bottom
Parameters for options 1 and 20 or command
===>
F1=Help F3=Exit F4=Prompt F5=Refresh F9=Retrieve F12=Cancel
F13=Work with cluster menu
5. When you are ready to re-establish replication in the original direction from the preferred
primary to the preferred backup node, the next step is to vary off the independent ASP
on the backup node and re-attach the ASP session on the backup node using the
following command:
CHGASPSSN SSN(IASP1_MM) OPTION(*REATTACH)
Bottom
Parameters for options 1 and 20 or command
===>
F1=Help F3=Exit F4=Prompt F5=Refresh F9=Retrieve F12=Cancel
F13=Work with cluster menu
In the example above, the REATTACH option is run on the backup node, which means that
production-independent ASP data will copied to the backup independent ASP. This direction
is still consistent with the cluster resource group status.
3. You can now run the REATTACH option in the proper way, which is from the current
production site to the current backup site (that is, on the current backup node)
(Example 13-27) with CHGASPSSN.
Note: Do not use CHGASPSSN for asymmetrical Global Mirror. Instead use the DS8000
failbackpprc command. Refer to specific operations for asymmetrical Global Mirror.
Reversing FlashCopy
Note: To make the FlashCopy reversible you need to start the session with the options
FLASHTYPE(*COPY) and PERSISTENT(*YES).
Reverse operation of the FlashCopy causes the source volume data to be overwritten by
target volume data. This operation might be helpful if you want to remove the changes done
to the source IASP since the FlashCopy was started. In this example we use the same
environment as in “Environment overview” on page 295. The only changes we made was
removing system names in the ASP copy description, so the IASP target copy will not be
accessed by any other system. To do this we used the following command:
CHGASPCPYD ASPCPY(ASPCPYD2) LOCATION(*NONE)
In addition, we used target disks not included in any volume group on the D8000.
Copy Descriptions
ASP
device Name Role State Node
IASP3 ASPCPYD1 SOURCE AVAILABLE DEMOHA
IASP3 ASPCPYD2 TARGET UNKNOWN *NONE
Bottom
Press Enter to continue
Copy Descriptions
ASP
device Name Role State Node
IASP3 ASPCPYD2 SOURCE UNKNOWN *NONE
IASP3 ASPCPYD1 TARGET AVAILABLE DEMOHA
Bottom
Press Enter to continue
Incremental FlashCopy
When you have a FlashCopy target that is used for a longer period of time (for example, as a
source for your data warehouse), you might want to have the FlashCopy session active all the
time and do the incremental updates of the target IASP. To do this you need to configure the
ASP copy descriptions pointing to your source and target volumes and allowing the IASPs to
be used by the source and target nodes. The ASP copy descriptions that we use for this
example is the same as in “Environment overview” on page 295. To do a incremental
FlashCopy we need to take the following steps:
1. Quiesce the source IASP with the following command:
CHGASPACT ASPDEV(IASP3) OPTION(*SUSPEND) SSPTIMO(30)
2. Start the FlashCopy session with the PERSIST(*YES) option, as we do here:
STRASPSSN SSN(SESFC1) TYPE(*FLASHCOPY) ASPCPY((ASPCPYD1 ASPCPYD2))
FLASHTYPE(*NOCOPY) PERSISTENT(*YES)
3. Resume the source IASP operation with the following command:
CHGASPACT ASPDEV(IASP3) OPTION(*RESUME)
4. Vary on the target IASP on the target node (DEMOFC) and use it.
Any time that you want to update the content of the target IASP take the following steps:
1. Quiesce the source IASP with the following command:
CHGASPACT ASPDEV(IASP3) OPTION(*SUSPEND) SSPTIMO(30)
2. Do the FlashCopy increment operation with the following command:
CHGASPSSN SSN(SESFC1) OPTION(*INCR)
Specifying TRACK(*YES) enables the DS8000 to track the changes that are done to the
source volumes, and allows the reattach operation to be done faster. When the session is
detached, the target volumes are not usable. In this state you can do the system maintenance
of the target. In the DS8000, the FlashCopy session is not present.
When you want to bring the session back, you need to issue the following command:
CHGASPSSN SSN(SESFC1) OPTION(*REATTACH)
Environment overview
For this scenario, we need to create several cluster items. Cluster, cluster monitors, IASP,
and administrative domain setup are covered in Chapter 11, “Creating a PowerHA base
environment” on page 199. Creating a PoweHA base environment is common to all
the scenarios.
Table 14-1 and Figure 14-1 on page 373 describe and show our environment.
IASP name, number IASP1, 144 / IASP2, 145 IASP1, 144 / IASP2, 145
Takeover IP 10.0.0.1
SCSI adapter resource name DC07 / DC09 DC08 / DC09 DC07 or DC01
a. Volumes IDs do not need to be the same on the source and target of remote copy relationships.
Power 75 0
Power 75 0
D3 D4 D5 D6 D7 D8 D9 D1 0
D3 D4 D5 D6 D7 D8 D9 D10
ISL
0 4 1 5 2 6 3 7 8 12 9 13 10 14 1 15 16 20 17 21 18 2 19 23 24 82 25 29 26 30 27 31 0 4 1 5 2 6 3 7 8 12 9 13 10 14 1 15 16 20 17 21 18 2 19 23 24 28 25 29 26 30 27 31
master
Syst emSt orage
FlashCopy
FlashCopy GlobalMirror
Syst emSt orage
Syst emSt orage
target
target
DS5300
DS5300
System Storage
System Storage
Generation of the SSH key pair is done on IBM i from QSHELL. Example 14-1 shows the
command to do this and a list of the files created.
Note: The ssh key pair generated here and used by PowerHA is in OpenSSH key format.
It cannot be used by PuTTY, as PuTTY expects SSH2 keys.
You then have to import the id_rsa.pub file as a key into a user on your SVCs. This user
must have administrator role to perform the functions used by PowerHA. Transfer the file to
your PC and import it to the SVC user (Figure 14-2).
Make sure to distribute the id_rsa file to all nodes in your cluster to the same directory.
IBM_2145:ctcsvcclu2:admin>svcinfo lsclustercandidate
id configured name
0000020065005ADE no ctcsvcclu1
2. After the SVC inter-cluster zoning has been configured, we create SVC cluster
partnerships between the two clusters using svctask mkpartnership on each cluster
(Example 14-3). Note that when the partnership has been configured only on one cluster it
is in a partially_configured state.
IBM_2145:ctcsvcclu1:admin>svcinfo lscluster
id name location partnership bandwidth id_alias
0000020065005ADE ctcsvcclu1 local
0000020065005ADE
0000020065FFFFFE ctcsvcclu2 remote fully_configured 200
0000020065FFFFFE
3. Before creating the remote copy volume relationships we create a remote copy
consistency group on each SVC cluster, which is required by PowerHA (Example 14-4).
IBM_2145:ctcsvcclu1:admin>svcinfo lsrcconsistgrp
id name master_cluster_id master_cluster_name aux_cluster_id aux_cluster_name primary state relationship_count copy_type
0 IASP1_MM 0000020065005ADE ctcsvcclu1 0000020065FFFFFE ctcsvcclu2 empty 0 empty_group
IBM_2145:ctcsvcclu1:admin>svctask mkrcrelationship -master CTCIHA9V_MM_V0 -aux CTCIHA9W_MM_V0 -cluster ctcsvcclu2 -consistgrp IASP1_MM
-name CTCIHA9_MM_V0
RC Relationship, id [15], successfully created
IBM_2145:ctcsvcclu1:admin>svctask mkrcrelationship -master CTCIHA9V_MM_V1 -aux CTCIHA9W_MM_V1 -cluster ctcsvcclu2 -consistgrp IASP1_MM
-name CTCIHA9_MM_V1
RC Relationship, id [16], successfully created
IBM_2145:ctcsvcclu1:admin>svctask mkrcrelationship -master CTCIHA9V_MM_V2 -aux CTCIHA9W_MM_V2 -cluster ctcsvcclu2 -consistgrp IASP1_MM
-name CTCIHA9_MM_V2
RC Relationship, id [17], successfully created
IBM_2145:ctcsvcclu1:admin>svctask mkrcrelationship -master CTCIHA9V_MM_V3 -aux CTCIHA9W_MM_V3 -cluster ctcsvcclu2 -consistgrp IASP1_MM
-name CTCIHA9_MM_V3
RC Relationship, id [18], successfully created
5. After the remote copy relationships have been created and added to the consistency
group we start the consistency group using svctask startrcconsistgrp (Example 14-6).
Note that after starting the consistency group its status changes from
inconsistent_stopped to inconsistent_copying.
6. We can now observe the progress of the SVC remote copy background synchronization
process using svcinfo lsrcrelationship (Example 14-7).
Note: As long as the SVC remote copy background synchronization progress has not
completed (that is, it has not reached consistent_synchronized), the remote copy
relationships should not be switched, as the data on the remote site is still inconsistent.
Bottom
F3=Exit F4=Prompt F5=Refresh F12=Cancel F13=How to use this display
F24=More keys
Figure 14-4 Add SVC ASP Copy Description
Unless you are on SVC/V7000 6.2 or later, the user name that you specify here has to be
admin. This profile has no relationship to the user that is used on the SVC. The actual user
is chosen based on the ssh key file pair only. With SVC6.2 you can either specify admin
as the user or the name of the SVC/V7000 user that the ssh keyfile actually belongs to.
The virtual disk range is the SVC volume IDs of the disks in your IASP.
5. Start the ASP session using STRSVCSSN (Figure 14-5).
Bottom
F3=Exit F4=Prompt F5=Refresh F12=Cancel F13=How to use this display
F24=More keys
Figure 14-5 Start SVC Session
Your IBM PowerHA SystemMirror i SVC/V7000 remote copy configuration is now done and
your IASP is highly available for planned and unplanned site switches, as described in 14.2,
“Managing IBM i SVC/V7000 Copy Services” on page 386.
To create a FlashCopy point-in-time copy for an IASP in the IBM i cluster device domain,
perform the following steps:
1. Create ASP copy descriptions for both the FlashCopy source and the target volumes.
2. Start an ASP session with type *FLASHCOPY, which links the copy descriptions and
creates the FlashCopy relationships on the SVC/V7000 storage system.
Next we describe these steps for an example scenario of configuring FlashCopy on a Metro
Mirror secondary site based on our SVC/V7000 environment (as shown in 14.1, “SVC/V7000
Copy Services” on page 372), taking the FlashCopy from the Metro Mirror secondary volumes.
Because we have not added our IBM i partition CTCIHA9X to the cluster and device domain,
we must add it as the third node to the cluster and device domain first before being able to
create an ASP copy description referencing it (Example 14-8). We use the IBM i partition as
the FlashCopy partition accessing the FlashCopy target volumes for doing a backup to
physical tape.
Example 14-8 Adding the FlashCopy node into the cluster and device domain
ADDCLUNODE CLUSTER(PWRHA_CLU) NODE(CTCIHA9X ('10.10.10.3'))
ADDDEVDMNE CLUSTER(PWRHA_CLU) DEVDMN(PWRHA_DMN) NODE(CTCIHA9X)
We also need to create a device description for the IASP on the FlashCopy target node
(CTCIHA9X in our example) using CRTDEVASP (Example 14-9). Make sure to specify the same
database name here that you specified when creating your original IASP.
Example 14-9 Creating the ASP device description on the FlashCopy target node
CRTDEVASP DEVD(IASP1) RSRCNAME(IASP1)
Bottom
F3=Exit F4=Prompt F5=Refresh F12=Cancel F13=How to use this display
F24=More keys
Figure 14-6 Add SVC/V7000 ASP Copy Description for FlashCopy target volumes
Note: For the target copy description that is to be used for FlashCopy, the cluster resource
group and cluster resource group site must be *NONE. The node identifier must be set to
the cluster node name that will own the target copy of the independent ASP.
Bottom
F3=Exit F4=Prompt F5=Refresh F12=Cancel F13=How to use this display
F24=More keys
Figure 14-7 Starting an SVC/V7000 FlashCopy session
Note: IBM PowerHA SystemMirror for i automatically creates the FlashCopy mappings
within the SVC/V7000 when starting an ASP session for type *FLASHCOPY. That is,
the FlashCopy mappings between the VDisks must not be created on the SVC/V7000
by a user before STRSVCSSN is executed.
STRSVCSSN allows the user to also create an incremental FlashCopy relationship and
specify a FlashCopy background copy rate and grain size. PowerHA requires the
FlashCopy relationships to be included in a consistency group that is by default newly
created by PowerHA on the SVC/V7000 when starting a FlashCopy session. Alternatively,
the user can specify the name for an existing FlashCopy consistency group. For further
information about these parameters see 7.4, “FlashCopy” on page 124. Information about
managing a FlashCopy environment with SVC/V7000 can be found in 14.2, “Managing
IBM i SVC/V7000 Copy Services” on page 386.
Example 14-10 CHGASPACT run from the FlashCopy target node for quiescing an IASP
PGM
RUNRMTCMD CMD('CHGASPACT ASPDEV(IASP1) +
OPTION(*SUSPEND) SSPTIMO(30)') +
RMTLOCNAME(CTCIHA9V *IP) RMTUSER(POWERHA) +
RMTPWD(REDB00K)
STRSVCSSN SSN(SVC_FLC) TYPE(*FLASHCOPY) +
ASPCPY((SVC_MM_T SVC_FLC_T))
RUNRMTCMD CMD('CHGASPACT ASPDEV(IASP1) +
OPTION(*RESUME)') RMTLOCNAME(CTCIHA9V +
*IP) RMTUSER(POWERHA) RMTPWD(REDB00K)
VRYCFG CFGOBJ(IASP1) CFGTYPE(*DEV) STATUS(*ON)
/* INSERT CALL OF YOUR BACKUP PROGRAMS HERE */
VRYCFG CFGOBJ(IASP1) CFGTYPE(*DEV) STATUS(*OFF)
ENDSVCSSN SSN(SVC_FLC)
ENDPGM
Notes: STRSVCSSN and ENDSVCSSN must be executed from the IBM i cluster node that
owns the target copy of the IASP.
The *DETACH and *REATTACH operations are not supported for a SVC/V7000
FlashCopy session.
Bottom
Copy Descriptions
ASP Copy
device Role Node progress Storage state
IASP1 SOURCE CTCIHA9W 0 Copying
TARGET CTCIHA9X
Press Enter to continue
2&fccstgrp2&empty
09/27/2011 17:31:39 950F29B4CA3B3001 <INFO> : YaspSVCSession : startFlashCopy : FlashCopy session SVC_FLC started successfully
09/27/2011 17:31:39 950F29B4CA3B3001 <INFO> : yaspSVCActionPgm : doSessionAction : Session action completed with return code: 1
09/27/2011 17:31:39 950F29B47F165001 <INFO> : YaspSession : iapEnlistReject : Start of IASP enlist reject: 6
144
09/27/2011 17:31:39 950F29B47F165001 <INFO> : YaspSession : iapEnlistReject : iaspEnlistReject succeeded
09/27/2011 17:31:39 950F29B47F165001 <INFO> : YaspSession : waitForUnitsToEnlist : Start of wait for units to enlist for session: SVC_FLC
09/27/2011 17:31:47 950F29B47F165001 <INFO> : YaspSession : waitForUnitsToEnlist : Disk units for all ASPs in session: SVC_FLC have
enlisted
09/27/2011 17:31:47 950F29B47F165001 <INFO> : YaspSession : resetMultipath : Start of reset multipath: 144
09/27/2011 17:31:47 950F29B47F165001 <INFO> : YaspSession : resetMultipath : Reset multipath succeeded
09/27/2011 17:31:47 950F29B47F165001 <INFO> : YaspPluginSession : start : Plugin session start for ASP Session: SVC_FLC completed
successfully.
Bottom
Copy Descriptions
Bottom
ASP Copy
device Role Node progress Storage state
IASP1 SOURCE CTCIHA9V 100 Consistent sync
TARGET CTCIHA9W
Bottom
Press Enter to continue
Using CHGSVCSSN ... OPTION(*CHGATTR), only the switchover and failover reverse replication
parameters can be changed.
An SVC/V7000 remote copy ASP session can be suspended from the IBM i cluster node that
owns the primary or secondary copy of the IASP, using CHGSVCSSN ... OPTION(*SUSPEND),
which puts the SVC/7000 remote copy relationships in consistent_stopped state.
The backup node becomes ineligible while the remote copy replication is suspended or
resuming (Figure 14-10). That is, the CRG cannot be switched or failed over before the
remote copy status becomes consistent_synchronized again.
Bottom
Parameters for options 1 and 20 or command
===>
F1=Help F3=Exit F4=Prompt F5=Refresh F9=Retrieve F12=Cancel
F13=Work with cluster menu
Figure 14-10 Work with Recovery Domain panel after suspending an ASP session
A suspended SVC/V7000 Metro Mirror or Global Mirror ASP session (because data
consistency is ensured by using SVC/V7000 remote copy consistency groups) can also be
detached, which puts the remote copy relationship in idling state to allow access of the
secondary copy of the IASP from the IBM i backup node.
You can then detach the session from your backup node using the command shown in
Example 14-13.
The reattach again has to be done from the backup node. Any changes made to the
secondary copy IASP while detached will be overwritten by the data from the primary ASP
copy. Vary off the IASP on the backup node and issue the command shown in
Example 14-14.
By default, a message in the QSYSOPR message queue tells you which node will be acting
as the primary node after the reattach and waits for confirmation before starting remote copy
services on the SVC/V7000 again. If you want to avoid this behavior, you can specify the
SNDINQMSG(*NO) parameter.
Session . . . . . . . . . . . . . . . . . . : SVC_MM
Type . . . . . . . . . . . . . . . . . . . : *METROMIR
Copy Descriptions
ASP Copy
device Role Node progress Storage state
IASP1 SOURCE CTCIHA9V 100 Consistent sync
TARGET CTCIHA9W
Figure 14-11 Displaying an ASP session for SVC Metro Mirror replication
Each scenario requires different failover/recovery actions, which we describe in further detail
in the following sections.
For an unplanned automatic failover event, a CPABB02 inquiry message is sent to the cluster
or CRG message queue on the backup node if a failover message queue is defined for either
the cluster or the CRG. If no failover message queue is defined, then the failover starts
immediately without any message being posted.
Bottom
Copy Descriptions
ASP Copy
device Role Node progress Storage state
IASP1 SOURCE CTCIHA9W 100 Idling
TARGET CTCIHA9V
Bottom
Press Enter to continue
After recovery of the preferred primary node, including restart of its cluster services, the
detached ASP session should be re-attached to resume remote copy replication and make
the IASP highly available again. There are options for re-attaching the ASP session:
To resume remote copy data replication from the secondary site back to the primary use
CHGSVCSSN ... OPTION(*REATTACH) on the preferred primary site node, which is the
current backup node. A planned switchback to the preferred primary node can then be
done using CHGCRGPRI.
Alternatively, if instead the data updates performed on the secondary site while the
primary site node was not available will be discarded, the direction of remote copy data
replication can be changed from the preferred primary node to the preferred backup node
by ending the CRG, changing the recovery domain primary/backup node roles via CHGCRG
before reattaching the session via CHGSVCSSN ... OPTION(*REATTACH) to resume
replication between the preferred primary and preferred backup node site.
In this case the user should carefully consider whether it is more appropriate to try to recover
from the access loss condition before invoking a failover to the backup cluster node, which
requires the following manual actions. If the primary node is still responsive, the ASP session
can be detached as follows to stop the remote copy relationships by allowing access to the
secondary volumes so that IASP can be varied on at the backup node:
CHGASPSSN SSN(<ASP_session>) OPTION(*DETACH)
The *DETACH operation for a failed IASP implies that the node roles still need to be changed
via ENDCRG and CHGCUR RCYDMNACT(*CHGCUR).
Otherwise, if the primary node is no longer responsive, a cluster partition condition and
failover can be enforced from the backup node as follows:
CHGCLUNODE CLUSTER(<cluster_name>) NODE(<primary_node_name>) OPTION(*CHGSTS)
Note: If CHGCLUNODE first fails with message CPFBB89, retry the command a short time
later after the cluster changed to a partition condition.
A CHGCLUNODE triggers a cluster failover without a failover inquiry message and still requires
the user to vary on the IASP on the new primary node and start the takeover IP interface.
After recovery of the preferred primary node, including a restart of its cluster services and of
the device CRG, the ASP session should be reattached on the preferred primary node, which
is the current backup node to resume replication. A switchback to the preferred primary node
can then be done using CHGCRGPRI.
A cluster partition condition with a failed cluster node indicated by message ID CPFBB20
requires manual actions to invoke a cluster failover, for example, by setting the primary
cluster node from partition to failed status using the following CL command:
CHGCLUNODE CLUSTER(<cluster_name>) NODE(<primary_node_name>) OPTION(*CHGSTS)
After this CHGCLUNODE command, cluster node roles are changed and remote copy
relationships are stopped so that the IASP can be manually varied on at the new primary
node on the backup site.
Bottom
Copy Descriptions
Bottom
ASP Copy
device Role Node progress Storage state
IASP1 SOURCE CTCIHA9W 0 Copying
TARGET CTCIHA9X
Bottom
Press Enter to continue
When using FlashCopy with the background copy (that is, copy rate > 0), the background
copy progress is reported in the copy progress information with a storage state of copying
when the background copy has not finished yet or copied when the background copy has
completed, as also indicated by a copy progress of 100.
Using CHGSVCSSN ... OPTION(*CHGATTR), only the copy rate and cleaning rate parameters of
a FlashCopy ASP session can be changed. This operation can be executed from either the
IBM i cluster node that owns the source copy or the target copy of the IASP.
Using incremental FlashCopy requires that the ASP session for the initial FlashCopy is
created with the incremental flash option set to *YES using STRSVCSSN ... TYPE(*FLASHCOPY)
INCR(*YES). A background copy rate greater than 0 should be specified when starting the
ASP session for the initial FlashCopy because any subsequent incremental FlashCopy
operations can only be performed using CHGSVCSSN with the *INCR option after the initial
background copy has finished.
As with any other FlashCopy operation, the FlashCopy incremental operation must also be
executed from the IBM i cluster node that owns the target copy of the independent IASP.
Bottom
Copy Descriptions
Bottom
ASP Copy
device Role Node progress Storage state
IASP1 SOURCE CTCIHA9W 100 Idle or copied
TARGET CTCIHA9X
Bottom
Press Enter to continue
If the background copy has not been enabled or finished, the incremental FlashCopy
operation is rejected with the following CMMVC5907E message:
The FlashCopy mapping or consistency group was not started because the mapping or
consistency group is already in the copying state.
In this case the user should either wait for an existing background copy process to complete,
which can be checked from the copy progress information displayed by DSPSVCSSN, or enable
the background copy by using CHGSVCSSN ... OPTION(*CHGATTR) CPYRATE(...).
For further details about PowerHA functions and issues, you might also want to consult the
“IBM Software Knowledge Base” section related to high availability at the following URL:
http://www-912.ibm.com/s_dir/slkbase.NSF/wHighAv?OpenView&view=wHighAv
To avoid product commands from becoming unusable in case of processor activation related
to temporary Capacity On Demand feature usage, make sure to apply PTF SI41735.
15.1.2 Requirements
When looking at the cluster properties with the GUI, there is an option to check the
requirements. Make sure to use this option and apply all of them. As shown in Figure 11-10
on page 207, these are the most common requirements to be applied:
System value QRETSVRSEC must be set to 1. This value is necessary for the
administrative domain to retrieve user profiles passwords.
Note: Considerations apply regarding this system value, which are described in the
knowledge base document number 518525704 Admin Domain: What to Consider For
User Profiles When Implementing, at the following URL:
http://www-912.ibm.com/s_dir/slkbase.NSF/cd034e7d72c8b65c862570130058d69e/92
eb712dc067c1558625757b006a2a55?OpenDocument
The QGPL/QBATCH job queue entry for QSYS/QBATCH must have *NOMAX or a value
greater than 1. This value is necessary because most of the cluster jobs are submitted to
this job queue, and they must not be prevented from running by user jobs.
QIBM_PWRDWNSYS_CONFIRM and QIBM_ENDSYS_CONFIRM environment
variables should be set to *YES.
Caution: Any IASP existing on a system before adding this system to a device domain
must be deleted.
15.1.4 Communications
The inetd service must be started on all cluster nodes.
When using advanced node failure detection, the HMC or VIOS should be able to send
appropriate information to cluster nodes in case of partition failure. There is no specific
configuration step for this item, but the name or IP address of the HMC or VIOS should be
available. Therefore, this communication takes the IP routes available when needed. You
might want to make sure that no network device failure could have an impact on this
communication.
If there is a firewall between the HMC or VIOS and the cluster nodes, make sure that the
flow is authorized. Advanced node failure detection makes use of the CIMOM services,
which listen on TCP 5989 (unencrypted communication) and 5990 (ssl encrypted
communication) ports. Flow must be authorized in both ways.
Communication with DS8000
When using DS8000 replication, the cluster nodes need to establish IP communication
with the DS8000 to check the replication status and start the appropriate actions, such as
failover or failback. There is no specific configuration step for this item, but the name or IP
address of the DS8000 should be available. Therefore, this communication will take the IP
routes available when needed. You might want to make sure that no network devices
failure can have an impact on this communication.
If there is a firewall between the cluster nodes and the DS8000, make sure that the flow is
authorized. DS8000 listens on the TCP 1750 port. The flow must be authorized from the
nodes to the DS8000.
15.2 Journaling
It is difficult to achieve both Recovery Time Objective (RPO) and Recovery Point Objective
(RTO) with an hardware replication solution such as IBM PowerHA SystemMirror for i without
looking seriously to a journaling setup. This solution relies on disk replication. Therefore, it
copies, from source disks to target disks, only data that exists on disks. Everything that is still
in main memory and that has not been yet written to disk cannot be replicated. It is not a
matter for planned switches, for which you will use varying off the IASP, which flushes
memory content to disks. It becomes a matter for unplanned switches, which can occur at any
time, for any reason, and for which we have to apply techniques to prevent loosing data and
to reduce recovery time.
Let us see an example. Assume that an application makes use of a data area to keep track of
the next customer number of a database file to record registration information for the new
customer to be assigned and of an IFS file to capture a scanned image of the customer’s
This lack of consistency on disks drives clearly affects the ability to reach planned RPO.
Some objects, like database files, might also need time-consuming internal consistency
checks and recovery at IASP vary-on time. Detecting and correcting them affects planned
RTO also.
Using journal is the way to go for achieving a consistent state and reducing recovery times as
much as possible. Journal protection should be enabled for all critical recovery point
objectives, data areas, database files, data queues, and IFS files. Using journal does not
require an update to the application code. Starting, ending, and managing journal operations
are to be done outside the application. The best protection for data consistency is achieved
by changing the application to take care of transaction consistency with commitment control.
The application decides when, from a functional standpoint, any transaction is finished and
then it performs a commit. If an outage occurs during commitment control cycles, the system
ensures, at IASP vary-on time, that data is consistent regarding the transactions. It undoes
any update that is not committed.
Journaling has never been mandatory with the IBM i operating system. Therefore, numerous
users have never tried to implement it, even when using hardware replication solutions. Note
that software replication solutions are based on journaling, so considerations differ for them.
Note: It is a user responsibility to take care of journaling. IBM PowerHA SystemMirror for i
has no option to help you with it.
However, in IBM i release after release, major improvements have been done on this subject:
When using internal drives for IASP, write cache performance is a key factor for journal
performance. For more information about this topic, refer to Journaling - Configuring for
your Fair Share of Write Cache, TIPS0653, at:
http://publib-b.boulder.ibm.com/abstracts/tips0653.html
If you decide to use a private ASP for receivers, do not skimp on the quantity of disk arms
available for use by the journal. For more information about this topic, refer to Journaling -
User ASPs Versus the System ASP, TIPS0602, at:
http://publib-b.boulder.ibm.com/abstracts/tips0602.html
Note: Do not forget that a private (user) ASP can be created inside an IASP as a
secondary ASP type and coexist with a primary type ASP.
Note: PTF MF51614 for IBM i 7.1 apply result is that journal receivers are spread
across all the disks just like any other object type, via normal storage management
technique. The journal threshold is no longer used to determine the number of disks
required for journaling.
Make sure that your journal is employing the most recent defaults. Many journals created
prior to release IBM i 5.4 might still be locked into old settings, and these old settings
might be thwarting performance. One of the easiest ways to ensure that you remain in
lock-step with the best journal settings is to let the system apply its own latest defaults.
You can help ensure that such settings are employed by specifying the
RCVSIZOPT(*SYSDFT) parameter on CHGJRN.
Consider installing and enabling the journal caching feature if journal performance
(especially during batch jobs) is a major concern in your shop. Make sure to understand
possible impacts on missing receiver writes in case of an outage. For more information
about this topic, refer to Journal Caching: Understanding the Risk of Data Loss,
TIPS0627, at:
http://www.redbooks.ibm.com/abstracts/tips0627.html
The more actively being-modified objects (such as open files) that you have associated
with a journal, the higher you might want to set your journal recovery count. Failure to do
so slows your runtime performance and increases your housekeeping overhead on your
production system without adding much benefit to your RTO. Increasing this value might
make good sense, but do not get carried away. For more information about this topic, refer
to The Journal Recovery Count: Making It Count on IBM i5/OS, TIPS0625, at:
http://www.redbooks.ibm.com/abstracts/tips0625.html
If you have physical files that employ a force-write-ratio (the FRCRATIO setting) and those
same files are also journaled, disable the force-write-ratio. Using both a force-write-ratio
and journal protection for the same file yields no extra recovery/survivability benefit and
only slows down your application. It is like paying for two health insurance policies when
one will suffice. The journal protection approach is the more efficient choice.
More in-depth treatments of a number of these best practices can be found in Striving for
Optimal Journal Performance on DB2 Universal Database for iSeries, SG24-6286, at:
http://www.redbooks.ibm.com/abstracts/sg246286.html
There are also numerous technotes that are probably more recent than the publication above,
which can be found on the IBM Redbooks publication website.
You might want to try this query URL to find information about Journaling and IBM i:
http://publib-b.boulder.ibm.com/Redbooks.nsf/searchdomain?SearchView&query=[subjec
ts]=AS400+and+Journaling&SearchOrder=1&category=systemi
Bottom
F3=Exit F4=Prompt F5=Refresh F12=Cancel F13=How to use this display
F24=More keys
Figure 15-1 STRJRNLIB
By using this command, all objects created in, moved to, and restored in the library will get
automatic journaling start to the same journal, with the same settings. There is no longer a
need to take care of them. However, you have to take care of those objects that you do not
want to journal. End journaling must be run for them.
Note: This command does not take care of existing objects in the library. It only applies to
changes to the library.
Journaling information:
Currently journaled . . . . . . . . : YES
Current or last journal . . . . . . : MYJOURNAL
Library . . . . . . . . . . . . . : MYLIB
Journal images . . . . . . . . . . . : *AFTER
Omit journal entry . . . . . . . . . : *NONE
New objects inherit journaling . . . : *YES
Inherit rules overridden . . . . . . : NO
Journaling last started date/time . : 09/20/11 17:00:25
Starting receiver for apply . . . . :
Library . . . . . . . . . . . . . :
Library ASP device . . . . . . . . :
Library ASP group . . . . . . . . :
Bottom
F3=Exit F10=Display inherit rules F12=Cancel Enter=Continue
Figure 15-2 Journaling information for MYLIB library
Considerations apply with another method to automatically start journaling, introduced in i5/OS
V5R4, with QDFTJRN data area. Form more information about these considerations and
about STRJRNLIB, refer to Journaling at Object Creation with i5/OS V6R1M0, TIPS0662, at:
http://www.redbooks.ibm.com/abstracts/tips0662.html
Bottom
F3=Exit F4=Prompt F5=Refresh F12=Cancel F13=How to use this display
F24=More keys
Figure 15-3 STRJRNPF command
Therefore, for an existing library with four commands, it is possible to start journaling for all
existing applicable objects and for all changes that apply in the future. Example 15-1 shows
an example with library name MYLIB.
The nice thing about the pseudo journal tool is that it not only helps answer these questions, it
does so without having a high impact on your system as it performs the analysis. Better yet, it
produces a customized analysis of projected additional disk traffic, tuned to your particular
application and its database reference pattern.
More information regarding the pseudo journal tool along with software to download and a
tutorial can be found on the database tools website:
http://www-03.ibm.com/systems/i/software/db2/journalperfutilities.html
At a minimum, we recommend a switchover every six months. This way, you can run half a
year on one site, switch over to the other site, and start again six months later. A six month
switch frequency is usually enough to make sure that the hardware is ready to handle
production workload when needed. Also make sure that all people are trained to handle an
unplanned failover. Running production on a six-month time window on the same node
makes you comfortable with events that occur with an higher frequency than the week or the
month.
Note: Cluster resource groups with an inactive status are not handled by the
failover procedure.
Normally, if there is an issue with the replication, the recovery domain node is ineligible
instead of active, which prevents you from switching. Using DSPASPSSN or DSPSVCSSN gives you
the opportunity to double-check the replication status, while using DSPCRGINF allows you to
check the recovery domain node status.
Messages are sent to the cluster message queue, which can be monitored to fix related issues.
Fixing this event allows the related recovery node status to change from ineligible to active.
Make sure that the interactive job running CHGCRGPRI is not using the IASP anymore because
though the command succeeds, the interactive job gets disconnected otherwise.
When using a CRG or cluster message queue, the cluster parameters Failover wait time and
Failover default action can be used to either set up an automatic cluster failover or require a
user confirmation for the failover to proceed or be cancelled. If both a CRG and a cluster
message queue have been configured, the cluster message queue takes precedence over
the corresponding CRG settings.
Note: In any case, clustering varies off the IASP on the current production node first when
it detects a failover condition before issuing the failover message.
This UID and GID are used when looking at object ownership. To reduce the amount of
internal processing that occurs during a switchover, where the IASP is made unavailable on
one system and then made available on a different system, the UID and GID values should be
synchronized across the recovery domain of the device cluster resource group. If this is not
the case, then each object owned by a user profile with a non-matching UID needs to be
accessed, and the UID needs to be changed as part of the vary-on process of the IASP after
Access paths are not available while they are being rebuilt. Therefore, if there are specific
logical files that an application requires to be valid shortly after a vary-on is complete, the user
should consider specifying RECOVER(*IPL) to avoid a situation in which jobs will try to use
the them before they are valid.
Another option to consider is the journaling of access paths so that they are recovered from
the journal during a vary-on and do not need to be rebuilt. Access paths are journaled
implicitly by SMAPP, as discussed in 15.5.4, “System Managed Access Path Protection” on
page 410. To ensure that specific, critical access paths are journaled, they should be explicitly
journaled.
When the system rebuilds an access path after IPL time, you can use EDTRBDAP to modify
rebuild sequences to select those access paths that need to be rebuilt now. But, by default,
this command applies only to SYSBASE. To allow it to work with independent ASP access
paths, do the following:
1. Enter the following command, where YY is the IASP number:
CRTDTAARA DTAARA(QTEMP/QDBIASPEDT) TYPE(*DEC) LEN(4) VALUE(YY)
2. Enter EDTRBDAP when the iASP becomes ACTIVE or AVAILABLE. It might be necessary to
invoke the command multiple times while the iASP is being varied on and is ACTIVE. A
CPF325C message will be sent until the command is allowed to be used.
3. Enter the following command to delete the data area:
DLTDTAARA DTAARA(QTEMP/QDBIASPEDT)
SMAPP applies to the IASP vary-on (or IPL for the SYSBASE) step, which is responsible for
rebuilding access paths in case of damages after an outage. For access paths (or indexes in
the SQL world) dependant on large physical files (or tables in the SQL world), it might take a
considerable amount of time, which can be several hours. To avoid rebuilding the access
paths, SMAPP uses existing journals for access path updates behind-the-scene recording,
Access paths update recording occurs automatically at each ASP level (independent or not,
and including SYSBASE) depending on the following parameters:
Access path recovery time target for each ASP.
Access path recovery estimate for each ASP. Each access path for which the estimated
rebuild time is higher than the target will be protected by SMAPP.
SMAPP effect the overall system performance. The lower the target recovery time that you
specify for access paths, the greater this effect can be. Typically, the effect is not noticeable,
unless the processor is nearing its capacity.
In the Figure 15-4, you can see an example of an existing independent ASP installation.
There is no target for any Independent ASP. Therefore, the system, by itself, has started
access path journaling to achieve the overall system target, which is set to 50 minutes. Using
the F14 key, we can see which access paths are currently protected (Figure 15-4).
For this example, the user should review the targets to specify a better recovery time target,
for example, by specifying *MIN, which means the lower possible recovery time.
Bottom
F3=Exit F5=Refresh F12=Cancel F13=Display not eligible access paths
F14=Display protected access paths F15=Display unprotected access paths
Figure 15-4 DSPRCYAP command result
To avoid some of these partition conditions, the HMC managing the production node server
or the VIOS managing the production node partition can send information about the status of
Make sure that you analyze the cause of any partition condition when deciding to either
resolve it or start a manual cluster failover. It could be due to a temporary network issue,
which can be easily corrected. As soon as the temporary failure is corrected, on a 15-minute
cycle the partition status is automatically solved and the node status changes from partition
back to active.
If you really need to start a failover use CHGCLUNODE to change the node status from partition
to failed.
See “Metro Mirror and Global Mirror failover and failback” on page 350, for examples of
these scenarios.
There are, however, a few additional steps that need to be considered if using a
hosted environment:
Although the network server storage space can reside in an IASP, the network server
description and network server configuration objects needed for iSCSI cannot. Therefore,
make sure to use the administrative domain to keep the network server descriptions
synchronized between all nodes in your cluster.
For hosted IBM i, AIX, or Linux partitions, virtual SCSI adapters have to be defined on the
production system as well as on the backup system to connect a network storage space to
the hosted LPAR.
For a planned switch, make sure to first end all activities in your hosted environment. Then
power down your hosted environment and end the network server description before
issuing CHGCRGPRI.
After a planned switch or a failover, vary on the network server description on the backup
system and start your hosted environment on the backup system.
The following white paper provides sample exit programs that can be used to automate
the vary-off and vary-on processes for network server descriptions:
http://www-03.ibm.com/systems/resources/systems_i_os_aix_pdf_linux_aix_xsm_final
.pdf
Perform a switchover test before moving the environment to production to make sure that the
setup does work on the backup site (that is, that you have done the necessary configuration
steps on the backup site correctly).
If you need to run continuous operations and are confident with the release upgrade, you can
do a so-called rolling upgrade. During this process (described in 15.9.1, “Performing a rolling
upgrade in a clustering environment” on page 414) either the production or the backup
system is available to the user. Be aware though that there are periods of time when you do
not have a backup system that you could fail over to if your current production system fails.
If you like to test your applications on one system using the new release while keeping the
application environment intact on the other system (so that in case of problems you can
simply move back to using this environment with the old version of the operating system), do
the release upgrade on the backup system after detaching the IASP.
The general order of steps when performing IBM i and PowerHA release upgrades is
as follows:
1. If you are using geographic mirroring, you can suspend it with tracking.
2. Upgrade IBM i and PowerHA on the current backup node and restart the backup
cluster node.
3. If you are using geographic mirroring and suspended it, resume it to synchronize the IASP
from the current production to the current backup node.
4. Switch from production to the upgraded backup node.
Independent ASP database conversion from one release to another occurs when the
IASP is first varied on.
5. If you are using geographic mirroring, you can suspend it with tracking.
6. Upgrade IBM i and PowerHA on the current backup node and restart the backup
cluster node.
7. If you are using geographic mirroring and suspended it, resume it to synchronize the IASP
from the current production to the current backup node.
8. When all cluster nodes are on the same release, update the PowerHA version
with CHGCLUVER.
Note: CHGCLUVER might need to be used twice to update the PowerHA version if you
skipped a release at the upgrade, for example, from V5R4 to i 7.1.
Note: CHGCLUVER might need to be used twice to update the PowerHA version by two if
you skipped a release at the upgrade, for example, from V5R4 to i 7.1.
http://www-912.ibm.com/s_dir/SLKBase.nsf/a9b1a92c9292c2818625773800456abc/ff759abf
1f796e8e8625715d00167c07?OpenDocument
However, you will have to deal with questions like “What was the system name for the last
backup tape for this file I need to restore now?” because saves to tapes can potentially run on
production and backup nodes.
BRMS network configuration can help you. Make sure to share the media information at the
library level. With this setting, to find the last backup for your file you can run the following
command on each system of your production node, which can run the save to tapes:
WRKMEDIBRM LIB(MYLIB) ASP(MYASP) FROMSYS(MYSYS)
To restore the desired object (assume that the last save was done on the MYSYS system),
set your job to use the appropriate IASP, then run the following command:
RSTOBJBRM OBJ(MYFILE) SAVLIB(MYLIB) DEV(*MEDCLS) SAVLVL(*CURRENT) FROMSYS(MYSYS)
There is another disadvantage with this usage. BRMS does not allow appending backups to
tape owned by another BRMS system, and retention calculation is done at the BRMS system
level. This means that you might need much more media than with the proposed
implementation discussed later.
In any case, when using a BRMS network, we recommend using a dedicated IP addresses
for BRMS.
In this specific case, when a save operation is scheduled to run on a system with a BRMS
name different from the system name, other BRMS network members must be aware of the
save operation. They need to know the current tape usage by the system running the save, in
place of the usual system and they need to update the BRMS database of the system running
the save.
When there is no save operation for the usual production system, the BRMS OTHER system
must connect to the BRMS PRODUCTION system to be aware of real-availability tapes, and it
must update the PRODUCTION BRMS database when it writes to available tapes.
Role
Role BACKUP
PRODUCTION
Po wer 7 5 0
D3 D4 D5 D6 D7 D8 D9 D1 0
Po wer 7 5 0
D3 D4 D5 D6 D7 D8 D9 D1 0
BRMS network
Po wer 7 5 0
D3 D4 D5 D6 D7 D8 D9 D1 0
OTHER
Figure 15-6 BRMS network when no save operation is running and roles are the preferred ones
Role
Role BACKUP
PRODUCTION
Po wer 7 5 0
D3 D4 D5 D6 D7 D8 D9 D1 0
Po wer 7 5 0
D3 D4 D5 D6 D7 D8 D9 D1 0
BRMS network
Po wer 7 5 0
D3 D4 D5 D6 D7 D8 D9 D1 0
OTHER
Tape Library
Figure 15-7 BRMS network when running a save-to-tape operation on the backup node and the roles are preferred ones
Group . . . . . . . . . . : IASP1
Default activity . . . . . *BKUPCY
Text . . . . . . . . . . . *NONE
Bottom
F3=Exit F5=Refresh F10=Change item F11=Display exits
F12=Cancel F14=Display client omit status F24=More keys
Figure 15-8 BRMS control group for saving IASP1 objects only
Note: SYSBASE objects can also be included in this backup if they are included in the
administrative domain and, therefore, are identical on both PRODUCTION and
BACKUP nodes.
7. On the BACKUP system, we prepare the comeback for the BRMS environment:
a. End the Q1ABRMNET subsystem.
b. End the PRODUCTION BRMS IP interface.
c. Delete the QUSRBRM/Q1ABRMSYS data area.
8. On the BACKUP system, we need to send the BRMS database to the
PRODUCTION system:
a. Save the QUSRBRM library to a save file.
b. Send it to the PRODUCTION system.
9. On the PRODUCTION system, we need to restore the BRMS database updated with the
save-o-tape operation run on the BACKUP system. If any update has been done to the
PRODUCTION BRMS database before this point, it is lost. Take these steps:
a. Clear the QUSRBRM library.
b. Restore objects from the save done in step 8.
10.On the PRODUCTION system, start the BRMS IP interface and the Q1ABRMNET
subsystem. Any BRMS save and tape information regarding IASP is now on the
PRODUCTION system, and from now on this one is used within the BRMS network by
other members.
Note: BRMS can have only one system name at any time. This means that you cannot run
several saves to tapes at the same time for IASP copies from several nodes. You can still
use a single node for saves to tape of IASP copies from several nodes, but they must be
done in a serial way. You have to run the steps described in this section for each production
system, one after the other.
15.10.3 Run production IASP copy saves to tapes after a roles switch
The same concern exists after a role switch (that is, when the preferred backup node
becomes the production node and the preferred production node becomes that backup node,
but they keep their system name). You will probably want to run the save-to-tape operations
on the actual backup node.
Role
Role BACKUP
Po wer 7 5 0
D3 D4 D5 D6 D7 D8 D9 D1 0
PRODUCTION
Po wer 7 5 0
D3 D4 D5 D6 D7 D8 D9 D1 0
BRMS network
Po wer 7 5 0
D3 D4 D5 D6 D7 D8 D9 D1 0
OTHER
Figure 15-9 BRMS network when no save operation is running and roles are switched ones
Role
Role BACKUP
Po wer 7 5 0
D3 D4 D5 D6 D7 D8 D9 D1 0
PRODUCTION
Po wer 7 5 0
D3 D4 D5 D6 D7 D8 D9 D1 0
BRMS network
Po wer 7 5 0
D3 D4 D5 D6 D7 D8 D9 D1 0
OTHER
Tape Library
Figure 15-10 BRMS network when running a save-to-tape operation on the backup node and the roles are switched ones
After switching roles, when there is no BRMS activity, BRMS names must be swapped, and
related IP interfaces as well:
On a production role node:
a. Start the BRMS production IP interface.
b. Change the BRMS system name to PRODUCTION with Q1AOLD API, as
shown above.
On backup role node:
a. Start the BRMS backup IP interface.
b. Change the BRMS system name to BACKUP with the Q1AOLD API, as shown above.
We decide on the backup system (that is, the one that runs the saves to tapes), which the
production system will look like it did it itself. Configuration steps are done either through the
BRMS GUI or with the QBRM/Q1AOLD API.
To add a specific system synchronization, change the system name to make it look like the
backup was done by this system and synchronize the reference date/time using this command:
CALL QBRM/Q1AOLD PARM('HSTUPDSYNC' '*ADD' ‘SYSTEM’ ‘NETWORK ID' 'IASPNAME'
'*CHGSYSNAM')
To add a specific system synchronization, keep the name of who did the backup and
synchronize the reference date/time:
CALL QBRM/Q1AOLD PARM('HSTUPDSYNC' '*ADD' ‘SYSTEM' ‘NETWORK ID' 'IASPNAME '
'*NORMAL')
Geographic mirroring
When replacing the backup system, including its disks, in an environment using geographic
mirroring with internal disks you cannot simply do a save and restore operation. Perform
these steps:
1. To preserve your old backup system in case migration to the new backup system fails,
perform a detach of the ASP session using CHGASPSSN OPTION(*DETACH).
2. Power down the old backup system.
3. Deconfigure geographic mirroring from the production system using either the GUI
interfaces or CFGGEOMIR. This results in an error message stating that the backup system
cannot be found. You can tell the system to ignore this status and to proceed with the
deconfiguration. Be aware that your production IASP needs to be varied off to perform
this step.
4. Start the new backup system. The system should have unconfigured drives available to
become the new backup IASP. Make sure that clustering works between the production
system and the new backup system. It might be necessary to remove and add the backup
cluster node to and from the cluster and recovery domain.
5. Configure geographic mirroring from the production system to the new backup system
using either the GUI interfaces or CFGGEOMIR. Be aware that your production IASP needs
to be varied off to perform this step.
6. When the configuration of geographic mirroring is finished, vary on the production IASP
and make sure that geographic mirroring is active. A full resynchronization is required.
Exchanging the production system without first deconfiguring geographic mirroring and
reconfiguring it afterwards is also not possible. Consider doing a CHGCRGPRI to switch over to
the backup system, and then follow the steps described above.
Make sure to change your SAN switch zoning for the new server WWPNs.
If a DS8000 or SVC/V7000 remote copy secondary storage system is not replaced but needs
to be powered-cycled for a disruptive maintenance action, the primary storage system keeps
track of the changes such that only a partial resynchronization of the out-of-sync tracks will be
required after the secondary storage system becomes operational again. Usually this
Note: If you are using the manual method of collecting data, you must collect it on all
nodes of your cluster.
Tip: IBM Support might instruct you to collect additional data not included in these
instructions, or they might require less data then listed here. Contact IBM Support with
your problem and you will get the instructions for your specific case.
Entry type:
Major code . . . . . . . . . . 0000 0000-FFFF
Minor code . . . . . . . . . . 0000 0000-FFFF
Starting:
Date . . . . . . . . . . . . . 09/20/11 MM/DD/YY
Time . . . . . . . . . . . . . 00:00:00 HH:MM:SS
Ending:
Date . . . . . . . . . . . . . 09/20/11 MM/DD/YY
Time . . . . . . . . . . . . . 00:00:00 HH:MM:SS
F3=Exit F12=Cancel
From:
Date . . . . . . . . 09/22/11 MM/DD/YY
Time . . . . . . . . 14:55:28 HH:MM:SS
To:
Date . . . . . . . . 09/29/11 MM/DD/YY
Time . . . . . . . . 14:55:28 HH:MM:SS
Device selection:
Option . . . . . . . . . . 1 1=Types, 2=Resource names
Device types or Resource names
*ALL *ALL...
c. The general instructions for how to run the tool can be found in “Instructions for running
the advanced analysis macro” on page 434.
In case you are not using QMGTOOLS, you need to collect XSM logs that can be found in
/QIBM/UserData/HASM/hads.
Option Command
FLIGHTLOG
ADDRESSINFO
ALTSTACK
BATTERYINFO
CLUSTERINFO
CONDITIONINFO
COUNTERINFO
DISABLEFLASHSYNC
DSTINFO
EXCEPTCHAIN
FINDFRAMES
FINDPTF
More...
F3=Exit F12=Cancel
If you do not find the macro name that you want to run on the list, you can run it by specifying
1 in the Options column in the first row (the empty one) and specifying the macro name in the
Command column.
Option Command
1 AAExample
FLIGHTLOG
ADDRESSINFO
ALTSTACK
BATTERYINFO
CLUSTERINFO
CONDITIONINFO
COUNTERINFO
DISABLEFLASHSYNC
DSTINFO
EXCEPTCHAIN
FINDFRAMES
FINDPTF
More...
F3=Exit F12=Cancel
Command . . . . : AAExample
Options . . . . . -all
Dump title . . . . . . . . . . .
F3=Exit F12=Cancel
If you have problems accessing the IBM Systems Director Navigator for i, make sure that you
have started the HTTP Server ADMIN domain. Use STRTCPSVR SERVER(*HTTP)
HTTPSVR(*ADMIN). Check the ADMIN and ADMIN1-4 jobs in QHTTPSVR subsystem.
Additional information about the problem can be found in the following directories:
/QIBM/UserData/HASM/logs
/QIBM/UserData/OS/OSGi/LWISysInst/admin2/lwi/logs
The PTFs listed on that website are related to the iDoctor functions. If you are not planing to
use QMPGTOOLS for iDoctor functionality, you do not need to install them. Check the PTFs
for high availability mentioned in 8.1.3, “Power Systems requirements” on page 135.
1
We based our example in this book on the tool version that is current at the time of writing. The tool might have
been updated since then.
Where SAVF_LIB is the library where you put the downloaded SAVF, and QMGTOOLvrm is
a SAVF name for your IBM i release.
To collect the data with QMGTOOLS it is enough to install the tool on one of the nodes only.
When you restore the QMGTOOLS library you can then use ADDLIBLE QMGTOOLS and
issue GO MG to access the tools main menu and then select option 1 to access the HA data
collection, or you can go straight to this menu with GO HASMNU.
(DMPCLU)
Bottom
F3=Exit F4=Prompt F5=Refresh F12=Cancel F13=How to use this display
F24=More keys
2 There is one menu option used by IBM Support (GO MG, opt.1, opt.4) that requires the 5799PTL to be installed.
Note: To successfully access all nodes, collect the data and get it on single node. User
profiles should be able to FTP successfully to and from every node. In case you cannot
use FTP among your nodes, you need to collect the data on each node separately, as
described in “Collecting the general data on a single node” on page 443.
DEMOPROD userid
DEMOHA userid
DEMOFC userid
F1=Continue F3=Exit
Figure 15-21 Specify same user and password for all nodes (optional)
Nodes Status
----- ------
Selection or command
===>
(DMPCLUINF)
Bottom
F3=Exit F4=Prompt F5=Refresh F12=Cancel F13=How to use this display
F24=More keys
Selection or command
===>
The file content might vary due to the development of QMPGTOOLS and your configuration.
Bottom
F3=Exit F4=Prompt F5=Refresh F12=Cancel F13=How to use this display
F24=More keys
As a result of this analysis, there will be a spoolfile created in your interactive job. In the file
you can find the problems that the tool has found and additional information about what to
collect to debug the problem further, if necessary.
If you have any suggestions for improvements that might help us improve this tool, you can
contact Benjamin Rabe ([email protected]) or the authors of this publication to have your
suggestions forwarded to the appropriate development team.
dscli> lsgmir 0
Date/Time: October 6, 2011 4:55:24 PM CEST IBM DSCLI Version: 7.6.10.530 DS:
IBM.1750-13ABGAA
SessionID MasterID ID State %Success CGtime
=======================================================================
0x02 IBM.1750-13ABGAA 02 Running 100 10/06/2011 15:32:12 CEST
dscli> showgmir 0
Date/Time: October 6, 2011 4:55:26 PM CEST IBM DSCLI Version: 7.6.10.530 DS:
IBM.1750-13ABGAA
ID IBM.1750-13ABGAA/02
Master Count 1
Master Session ID 0x02
Copy State Running
Fatal Reason Not Fatal
CG Interval Time (seconds) 0
Coord. Time (milliseconds) 50
Max CG Drain Time (seconds) 30
Current Time 10/06/2011 15:32:15 CEST
CG Time 10/06/2011 15:32:14 CEST
Successful CG Percentage 100
FlashCopy Sequence Number 0x4E8DADDE
Master ID IBM.1750-13ABGAA
Subordinate Count 0
Master/Subordinate Assoc -
Figure 15-29 lsgmir for Global Mirror
3. The FlashCopy status for the copied volumes can be checked with lsflash (Figure 15-30).
If you need additional help with the command you can use the -help option with any of them.
The list of available commands is shown in DSCLI when you issue help.
Modern IBM i high-availability solutions are built on IBM i cluster resource services, or more
simply clusters. A cluster is a collection or group of multiple systems that work together as a
single system from an application perspective. Clusters provide the underlying infrastructure
that allows resilient resources, such as data, devices, and applications, to be automatically or
manually switched between systems or nodes (partitions on a single frame or on multiple
frames). It provides failure detection and response, so that in the event of an outage, cluster
resource service responds accordingly, keeping your data safe and your business
operational. PowerHA SystemMirror for i offers complete end-to-end integrated solutions for
high availability and disaster recovery, with special focus on the application availability
through planned or unplanned outage events.
This appendix discusses two other options for providing data resiliency:
Full-system storage-based Copy Services solutions
Logical replication solutions
Boot from SAN makes it easier to bring up a system environment that has been copied using
Copy Services functions such as FlashCopy or remote Copy Services. During the restart of a
cloned environment, you no longer have to perform the Recover Remote Load Source Disk
Unit through Dedicated Service Tools (DSTs), thus reducing the time and overall steps
required to bring up a point-in-time system image after FlashCopy or remote Copy Services
functions have been completed.
Cloning IBM i
Cloning has been a concept for the IBM i platform since the introduction of boot from SAN
with i5/OS V5R3M5. Previously, to create a new system image, you had to perform a full
installation of the SLIC and IBM i. When cloning IBM i, you create an exact copy of the
existing IBM i system or partition. The copy can be attached to another System i/Power
Systems models, a separate LPAR, or, if the production system is powered off, the existing
partition or system. After the copy is created, you can use it for offline backup, system testing,
or migration.
Boot from SAN enables you to take advantage of some of the advanced features that are
available with IBM system storage. One of these functions is FlashCopy. It allows you to
perform a point-in-time instantaneous copy of the data held on a LUN or group of LUNs.
Therefore, when you have a system that only has SAN LUNs with no internal drives, you can
create a clone of your system.
Important: When we refer to a clone, we are referring to a copy of a system that only uses
SAN LUNs. Therefore, boot from SAN is a prerequisite for this.
DS8000
Figure A-1 Full system FlashCopy
To obtain a full system backup of IBM i with FlashCopy, either a system shutdown or, as of
IBM i 6.1, a quiesce function is supplied that flushes modified data from memory to disk.
FlashCopy only copies the data on the disk. The IBM i quiesce for Copy Services function
(CHGASPACT) introduced with 6.1 allows you to suspend all database I/O activity for
*SYSBAS and IASP devices before taking a FlashCopy system image, eliminating the
requirement to power down your system.
The main consideration with this solution is distance. The solution is limited by the distance
between the two sites. Synchronous replication needs sufficient bandwidth to prevent latency
in the I/O between the two sites. I/O latency can cause application performance problems.
With Global Mirror all the data on the production system is asynchronously transmitted to the
remote DS models. Asynchronous replication via Global Copy alone does not guarantee the
order of the writes, and the remote production copy will lose consistency quickly. To
guarantee data consistency, Global Mirror creates consistency groups at regular intervals, by
default, as fast as the environment and the available bandwidth allows. FlashCopy is used at
the remote site to save these consistency groups to ensure that a consistent set of data is
available at the remote site, which is only a few seconds behind the production site. That is,
when using Global Mirror a RPO of only a few seconds can be achieved normally without any
performance impact to the production site.
Figure A-2 shows an overview of a full system remote copy solution by using IBM
system storage.
SAN
DS8000 DS8000
Figure A-2 Full system remote Copy Services
This is an attractive solution because of the extreme distances that can be achieved with
Global Mirror. However, it requires a proper sizing of the replication link bandwidth to ensure
that the RPO targets can be achieved. Testing should be performed to ensure that the
resulting image is usable.
When you recover in the event of a failure, the IPL of your recovery system will always be an
abnormal IPL of IBM i on the remote site.
Note: Using IBM i journaling along with Metro Mirror or Global Mirror replication solutions
is highly recommended to ensure transaction consistency and faster recovery.
Logical
Apply
Replication
Management and Control Process
Software
Most logical replication solutions allow for additional features beyond object replication. For
example, you can achieve additional auditing capabilities, observe the replication status in
real time, automatically add newly created objects to those being replicated, and replicate a
subset of objects in a given library or directory.
With this solution in place, you can activate your production environment on the backup
server via a role-swap operation.
One characteristic of this solution category is that the backup database file is “live”. That is, it
can be accessed in real time for backup operations or for other read-only application types,
such as building reports.
The challenge with this solution category is the complexity that can be involved with setting up
and maintaining the environment. One of the fundamental challenges lies in not strictly
policing undisciplined modification of the live copies of objects residing on the backup server.
Failure to properly enforce such a discipline can lead to instances in which users and
programmers make changes against the live copy so that it no longer matches the production
copy. Should this happen, the primary and backup versions of your files are no longer
identical. There are new tools provided by the software replication providers that perform
periodic data validation to help detect this situation.
Another challenge associated with this approach is that objects that are not journaled must go
through a checkpoint, be saved, and then be sent separately to the backup server. Therefore,
the granularity of the real-time nature of the process might be limited to the granularity of the
largest object being replicated for a given operation.
For example, a program updates a record residing within a journaled file. As part of the same
operation, it also updates an object, such as a user space, that is not journaled. The backup
copy becomes completely consistent when the user space is entirely replicated to the backup
system. Practically speaking, if the primary system fails, and the user space object is not yet
fully replicated, a manual recovery process is required to reconcile the state of the
non-journaled user space to match the last valid operation whose data was completely
replicated. This is one of the reasons why the application state and the state of the replicated
data at the target box are inherently asynchronous.
Another possible challenge associated with this approach lies in the latency of the replication
process. This refers to the amount of lag time between the time at which changes are made
on the source system and the time at which those changes become available on the backup
system. Synchronous remote journal can mitigate this for the database. Regardless of the
transmission mechanism used, you must adequately project your transmission volume and
size your communication lines and speeds properly to help ensure that your environment can
manage replication volumes when they reach their peak. In a high-volume environment,
replay backlog and latency might be an issue on the target side even if your transmission
facilities are properly sized. Another issue can arise when the apply process on the target
system cannot keep up with the incoming data. The target system cannot be used until this
lag data is fully applied.
Table A-2 Comparison with data resiliency options (IT operating environment)
PowerHA Full-system Logical replication
SystemMIrror for i storage-based Copy
Services
Required bandwidth Data change in IASP Data change in full Journal changes
system
Use of journal Yes, best practice Yes, best practice Source of replicated
data
Recovery time IASP vary on time Abnormal IPL time Time of applying lag
objective plus switchover
overhead plus sync
check
The publications listed in this section are considered particularly suitable for a more detailed
discussion of the topics covered in this book.
IBM Redbooks
The following IBM Redbooks publications provide additional information about the topic in this
document. Note that some publications referenced in this list might be available in softcopy
only.
Striving for Optimal Journal Performance on DB2 Universal Database for iSeries,
SG24-6286
Journaling - User ASPs Versus the System ASP, TIPS0602
Journaling at object creation on DB2 for iSeries, TIPS0604
Journaling: Why Is My Logical File Journaled?, TIPS0677
Journaling - How Can It Contribute to Disk Usage Skew?, TIPS0603
Journaling · Journal Receiver Diet Tip 1: Eliminating Open and Close Journal Entries,
TIPS0607
Journaling - *RMVINTENT: The preferred fork in the road for heavy journal traffic,
TIPS0605
Journaling · Journal Receiver Diet tip 2: Consider using skinny headers, TIPS0654
You can search for, view, download or order these documents and other Redbooks,
Redpapers, Web Docs, draft and additional materials, at the following website:
ibm.com/redbooks
Online resources
These websites are also relevant as further information sources:
G N
Geographic mirroring 50 N_Port ID Virtualization (NPIV) 136
Global Copy 102 Name Space 15
Global Mirror 53, 99 name space 16
Global Mirror session 102 network attributes 30
Nocopy option 110
H
Hardware Management Console 83 O
Hardware Management Console (HMC) 77, 136 object creation 18
Heartbeat monitoring 37 ODBC 31
host based replication 51 OptiConnect connections 30
host connection 92 Outqueues 33
HSL loop 48
P
I Partial synchronization 71
IBM PowerVM 136 Password validation program 29
IBM Storwize V7000 54, 114 Peer CRG 42
idle or copied 127 POWER Hypervisor™ (PHYP) 43
Idling 121 POWER6 48
Inactive job message queue 29 prepared 128
Inconsistent copying 121 preparing 127
Inconsistent stopped 121 prestarted job 31
incremental FlashCopy 126 Primary disk pool 14
independent auxiliary storage pools 14 Primary node 39
indirection layer 125 Problem log filter 29
Integrated Virtualization Manager (IVM) 136 production copy 49, 69
IO enclosures 82 Program Request Pricing Quotation (PRPQ) 134
IP address 16
Q
J QALWUSRDMN 28
JDBC 31 QATNPGM 28
job descriptions 27 QBOOKPATH 28
Job queues 27 QCFGMSGQ 28
journal receivers 27 QCTLSBSD 28
Journals 27 QDFTJOBD 33
Index 463
464 PowerHA SystemMirror for IBM i Cookbook
PowerHA SystemMirror for IBM i
Cookbook
PowerHA SystemMirror for IBM i Cookbook
PowerHA SystemMirror for IBM i Cookbook
PowerHA SystemMirror for IBM i Cookbook
PowerHA SystemMirror for IBM i
Cookbook
PowerHA SystemMirror for IBM i
Cookbook
Back cover ®
PowerHA SystemMirror
for IBM i Cookbook
®
Take advantage of IBM PowerHA SystemMirror for i is the IBM high-availability disk-based
clustering solution for the IBM i 7.1 operating system. When combined INTERNATIONAL
PowerHA to configure
with IBM i clustering technology, PowerHA for i delivers a complete TECHNICAL
and manage high
high-availability and disaster-recovery solution for your business SUPPORT
availability applications running in the IBM System i environment. PowerHA for i ORGANIZATION
enables you to support high-availability capabilities with either native
Find guidance on disk storage or IBM DS8000 or DS6000 storage servers or IBM
planning and Storwize V7000 and SAN Volume Controllers.
implementing The latest release of IBM PowerHA SystemMirror for i delivers a
PowerHA brand-new web-based PowerHA graphical user interface that BUILDING TECHNICAL
effectively combines the solution-based and task-based activities for INFORMATION BASED ON
your HA environment, all in a single user interface. PRACTICAL EXPERIENCE
Benefit from the latest
PowerHA solution This IBM Redbooks® publication provides a broad understanding of IBM Redbooks are developed
enhancements PowerHA for i. This book is intended for all IBM i professionals who are by the IBM International
planning on implementing a PowerHA solution on IBM i. Technical Support
Organization. Experts from
IBM, Customers and Partners
from around the world create
timely technical information
based on realistic scenarios.
Specific recommendations
are provided to help you
implement IT solutions more
effectively in your
environment.