Clustering

Types of Clustering
Objectives
At the end of this module the student will
understand the following tasks and
concepts.
 What clustering is and why you would
want it
 Clustering options
 Differences between various types of
clustering; advantages and
disadvantages
 Factors to consider when choosing a
cluster type
What is a cluster?
 My definition
 Multiple systems performing a single
function
 Black box
Why Cluster?
 Performance
 Availability
 Recoverability
Features
 Speedup
 Faster response times
 Transactions finish faster
 Scaleup
 More work done
 More capacity, more concurrent
transactions
 Scalability
Single Node Scaling

Users
 Scales to
multiple CPUs
 Doesn’t scale
beyond one
node
 Multiple
single points Server
of failure

Database
Database
Cluster Definitions
 Shared Nothing (Federated)
 Replicated Site
 Shared Disk
 Failover
 Active/Passive
 Active/Active
 Shared Everything
Shared Nothing Cluster
 Only one CPU is connected to a disk
 May have shared memory
 MPP Systems are Shared Nothing
 Other vendors have “Shared
Nothing” clusters
Federated (Shared
Nothing) Cluster
 Distributed database
(separate database
on each machine)
 Data is spread across
nodes; each machine
has part of the data
 Function is spread 1. Got it?
across nodes
 Two-Phase Commit Good!
3.

Server Server
2.
Got it!

Database Database
Replicated System
 Data replicated at
the server
(network) level or
at the storage
(SAN) level
Passive Node
 Multiple copies of Active Node

the same
database Server level
 Most common Server
Replication
or
Server

implementation is Storage level

Active/Passive Replication
Database Database
 Failover between
nodes
Shared Disk Cluster
 Shared file system
 Multiple systems attached to the same
disk
 All nodes must have access to data
 Only one database instance; only one
node has “ownership” of the shared disk
 Synchronization between systems; If one
node fails, then the other takes over
Cluster Interconnect
 Most Shared Disk clusters require some form of
Cluster Interconnect
 Network – i.e. Gigabit Ethernet
 Specialized – i.e. Infiniband, Myrinet
 Most clusters implement a “heartbeat” between
cluster nodes to monitor node health
 Multiple nodes require a switch
 Usually separated from the LAN
 Some shared disk clusters implement a
“heartbeat” mechanism to a quorum disk via the
SAN in addition to/instead of network heartbeat
 Oracle RAC implements Cache Fusion across the
interconnect
 Extra network traffic increases the throughput
requirements
 UDP implementation requires a separate network
Failover Cluster
 One system is a standby system for
another
 Only one system doing work at a
time
 Pseudo-Shared Disk
 Limited scalability in active/passive
mode
Failover Clustering

Users

 Fault tolerant
systems;
highly
available
 Basic failover
clusters don’t Server Server
scale beyond
two nodes

Database
Database
Active/Passive vs.
Active/Active
 Both are failover only
 Active/Passive
 One node is active
 The other is passive until failover
 Active/Active
 Still uses active/passive technology
 2 separate databases
 One is active on node A and passive on node
B
 The second database is active on node B and
passive on node A.
 Separate applications and user connections to
each of the different databases
Active/Passive

Node A Node B

 Node A is active
 Node B is passive
until/unless Node A
fails
 Only one Oracle license
is required
Active/Passive

X
Node A Node B

If Node A fails …
Active/Passive

X
Node A Node B

 Node B becomes
active
 Node A is dead
(definitely passive!)
until repaired and
then “failed back”
if necessary.
Active/Active

Node A Node B  Application Group A

and User Group A are
Application A Application B active on Node A
User Group A User Group B
 Application Group B
Passive Fail- Passive Fail-
and User Group B are
over for B over for A
active on Node B

 Each node serves as

failover for the other.

 2 separate databases.
Both nodes are not
accessing the same
data at the same time.

 Oracle license required

on each node
Switchover vs. Failover
 Many cluster systems utilize the concept
of Service Groups
 Service Groups allow granular control of
individual software packages (i.e.
individual Oracle instances)
 An individual group can be manually
moved to another server without affecting
other service groups – a “switchover”
versus a “failover”
 Adds greater management flexibility
 N-to-1 Failover
Configuration
Node Node Node Node
A B C D
 Node D is a
Failover
dedicated
failover node for
Application A Application D Application G Failover G
failures on Node
A, B, and C

X
User Group A User Group D User Group G

Application B Application E Application H Failover H

User Group B User Group E User Group H
 Extends number
of active nodes
Application C Application F Application I Failover I
User Group C User Group F User Group I
 A problem is that
once the failed
node is available,
Failback the Service
Groups on Node D
(failover node)
must failback to
original server to
restore High
Availability
 N + 1 Failover
Configuration
Node Node Node Node
A B C D
 Node D is a
Failover
dedicated failover
node for failures
Application A Application D Application G Failover G
on Node A, B, and
C

X
User Group A User Group D User Group G

Application B Application E Application H Failover H

User Group B User Group E User Group H
 Extends number
of active nodes
Application C Application F Application I Failover I
User Group C User Group F User Group I
 Once Node C is
restored, it
becomes the
failover node,
leaving Node D in
production.
 N-to-N Failover
Configuration
Node Node Node Node
A B C D

Failover G Failover H Failover I  Node C fails,

and its Service
Application A Application D Application G Application J
Groups are re-

X
User Group A User Group D User Group G User Group J

Application B Application E Application H Application K

distributed
User Group B User Group E User Group H User Group K across
Application C Application F Application I Application L surviving
User Group C User Group F User Group I User Group L
nodes
 Optimal
solution for > 2
nodes
 Implemented
on third party
failover
Third Party Clusters
 Support for extended cluster nodes –
up to 32 nodes for vendor Clustering
 Supports N + 1 and N - N failover
clustering
 Integrated with hardware and/or
software replication for long distance
“clusters”
Clustering Solutions from
Oracle
 Oracle Failsafe
 Oracle Data Guard
 Advanced Replication
 Shared Nothing Cluster
 Oracle Parallel Server
 Real Application Clustering (RAC)
Failsafe
 MS Clustering Enabled
 Two servers one disk subsystem
 Switches in the event of a hardware
failure
 Requires recovery
Standby Database
 Copy of Database (usually remote)
 Kept up to date with Archive Logs
 Oracle 8i feature
 Oracle 9i-10g version of a standby
database is Data Guard
Oracle Data Guard
 Mirrored Server
 Physical Standby
 Archive Logs are applied to the remote database
 Switchover occurs in the event of a failure
 Logical Standby
 Log Miner technology is used to generate SQL
 Standby Database can also be used for read-only
reporting
 Advantages
 Safe from user failure
 Can be in different location
 No recovery required
Advanced Replication
 Uses Updatable-Snapshots
 Replicates to another system
 Systems stay in sync
Oracle Parallel Server
 Shared disk cluster product
 Loosely Coupled
 Scalable performance
 No downtime in the event of a
system failure
 Replaced by RAC in 9i
True Shared Disk Server
(RAC)
 ONE database
 Separate multiple
instances (processes &
memory)
Node A Node B
 All nodes can access
data simultaneously
 Shared Everything
Cluster
 Transparent Application
Failover
 Oracle license required
on each node
 Highest level of cluster
functionality
Factors to Consider for
Clustering
 Which do you need most?
 High Availability – Failover Clusters, Synchronous Replication,
Data Guard
 Performance scalability – Active/Active failover clusters, N-to-
N failover clusters
 Both – Oracle RAC
 Administration complexity
 Failover clusters – relatively low
 Oracle RAC – relatively high
 Substantially less complex for 10g RAC than 9i RAC
 Local or long distance?
 Local – Failover, RAC
 Remote – Federated database, Replication, Standby
database/Data Guard
 Oracle license costs
 Active/Passive failover clusters – active nodes only
 Active/Active failover clusters, RAC – per node
Review
 What type of commit is required for a
Federated (shared nothing) cluster?
 What is the difference in how the database
is kept up-to-date in Oracle Data Guard vs.
Advanced Replication?
 What is the difference between N-to-1
failover clusters and N + 1 failover
clusters?
 How many databases are there in an 8
node Oracle RAC cluster?
Summary
 Types of clusters:
 Shared Nothing Clusters
 Federated databases
 Replication
 Shared Disk Clusters
 Failover
 Oracle RAC
 Failover Clusters
 Active/Passive
 Active/Active
 N-to-1
 N+1
 N-to-N
 Shared Everything Clusters
 Oracle RAC
 Choosing a cluster type involves trade-offs in
functionality, costs, and administration
complexity