1

2003 EMC Corporation. All rights reserved. 11

LUN and RAID Group Expansion
2
2003 EMC Corporation. All rights reserved. 22
LUN Expansion with MetaLUNs
MetaLUN 11
HOST
LUN 11
R5, 4+1
LUN 12
R5, 4+1
LUN 13
R1/0, 4
LUN 14
R1
LUN 15
R5, 8+1
A metaLUN is a type of LUN whose maximum capacity can be the combined
capacities of all the LUNs that compose it. The metaLUN feature lets you
dynamically expand the capacity of a single LUN (base LUN) into a larger unit
called a metaLUN. You do this by adding LUNs to the base LUN. You can also
add LUNs to a metaLUN to further increase its capacity. The attached host sees
one large LUN and is unaware of the component LUNs as they are made private
when combined into a metaLUN. Like a LUN, a metaLUN can belong to a
Storage Group, and can participate in SnapView, MirrorView and SAN copy
sessions. MetaLUNs are supported only on CX-Series storage systems.
3
2003 EMC Corporation. All rights reserved. 33
Expansion by Concatenation
Base LUN Comp B
New
Capacity A
Comp A
New
Capacity B
A concatenate expansion creates a new metaLUN component that includes the
new LUNs and appends this component to the end of the existing LUN or
metaLUN. There is no re-striping of data between the original storage and the
new LUNs. The components may be different RAID types and different
capacities. The chief advantage of the concatenate operation is that it completes
immediately. While this is not the recommended long-term solution it does
provide for a fast, flexible expansion but may have a negative impact on
performance if LUNs of different RAID types are used.
4
2003 EMC Corporation. All rights reserved. 44
Expansion by Striping
FLU x
Unused
Capacity
metaLUN 0
Before:
Unused Capacity
After:
A stripe expansion takes the existing data on the LUN or metaLUN, and restripes
(redistributes) it across the existing LUNs and the new LUNs you are adding.
Components LUNs must be of the same RAID type and they should be the same
size. Excess blocks in each LUN (based on smallest LUN) are wasted space,
because they will be ignored by FLARE. The striped metaLUN has the advantage
of higher spindle counts for improved performance. The stripe expansion may
take a long time to complete. The process concatenates new storage space, then
re-stripes existing data across all spindles in the new space. The original space is
available during the entire operation and the additional space can be made
available after re-striping is complete. Striped expansion can take a significant
period of time to complete.
5
2003 EMC Corporation. All rights reserved. 55
MetaLUN Limits
CX600
CX400
CX200
Max number of
components
per metaLUN
Max number of
FLARE LUNs per
striped component
16
8
8
32
32
16
In a CX600 array, a metaLUN can consist of a maximum of 16 components and a
component can consist of a maximum of 32 striped FLARE LUNs.
In a CX400 array, a metaLUN can consist of a maximum of 8 components and a
component can consist of a maximum of 32 striped FLARE LUNs.
In a CX200 array, a metaLUN can consist of a maximum of 8 components and a
component can consist of a maximum of 16 striped FLARE LUNs.
6
2003 EMC Corporation. All rights reserved. 66
LUN Expansion Rules
Striped expansion: FLARE LUNs within a striped
component must be
the same RAID type
the same size (in blocks)
the same disk type (Fibre Channel vs. ATA)
Concatenated expansion: All components within
metaLUN must be
either all protected (1, 1/0, 3, 5) or all non-protected (0)
the same disk type (Fibre Channel vs. ATA)
Expansion LUNs must not be
consumed by another driver
private or transitioning
Striped expansion and concatenated expansion each have a unique set of rules.
When expanding a LUN into a striped metaLUN, all the LUNs participating in the
metaLUN component must be protected by exactly the same RAID scheme, be
exactly the same size, and must all be of the same disk type either Fibre Channel
or ATA.
When expanding a LUN into a concatenated metaLUN, the rules are less
restrictive. The LUNs participating in the meatLUN component must all be either
protected or unprotected, can be any size, and must be either all Fibre Channel
or all ATA.
The expansion LUNs must not be part of a snapshot, clone, mirror or SAN Copy
session and must not be part of a Storage Group. Also they are not allowed to be
private LUNs or in a transitioning state.
7
2003 EMC Corporation. All rights reserved. 77
Concatenate Expansion Walk-Through
To expand a LUN, right click the LUN to be used as the Base LUN and select
Expandfrom the menu. Remember that the LUN ID information of the Base
LUN will be inherited by the metaLUN.
LUN expansion is initiated through the use of a wizard. The wizard introduction
screen states the objective of the wizard and clicking Next will start the definition
process. At any point in the process clicking Cancel will abort.
8
2003 EMC Corporation. All rights reserved. 88
Select Expansion Type
Before any choices can be made for the component expansion FLARE LUNs, the
expansion type must be selected. Both options, Stripe Expansion or Concatenate
Expansion are briefly described and either type may be selected. This walk-
through will show a concatenated expansion.
9
2003 EMC Corporation. All rights reserved. 99
Select Unused LUNs
The wizard will present all unconsumed (unused) LUNs in the array without
regard to Storage Processor ownership. The only restriction to the presented list
is that the LUNs to be used for expansion must be protected (RAID 1, 1/0, 3, or
5) if the Base LUN is protected. Notice that LUNs of all sizes are listed.
10
2003 EMC Corporation. All rights reserved. 10 10
Expansion LUN Selected
Once expansion LUN(s) are selected the wizard will automatically calculate the
maximum user capacity. Click Next to proceed with the wizard.
11
2003 EMC Corporation. All rights reserved. 11 11
Specify MetaLUN Capacity
A very important consideration when expanding LUNs is the affect on the host file
systems. If the target host system OS can not handle dynamic changes to the
volume size, host contact with the data in the Base LUN may be interrupted.
The option of leaving the metaLUN presented to the host at the original Base
LUN size may be selected by clicking Current Capacity. Presenting the
Maximum Capacity or a user defined Other Capacity are also available. If
Other Capacity is selected, the value entered must be a whole number as
decimal values are not allowed. For example, if 1.5 GB is to be made available
the value must be entered as 1500 MB. For this walk-through, Maximum
Capacity has been selected. Click Next to proceed with wizard.
12
2003 EMC Corporation. All rights reserved. 12 12
Specify MetaLUN Settings
MetaLUN Name by default is inherited from the Base LUN.
The Default Owner Storage Processor is the SP that assumes ownership of the metaLUN when the
storage system is powered up and is inherited from the Base LUN and can be user set.
The Expansion Rate will set the relative priority of expansion activity in relation to host I/O. It may be set to
low, medium, high, or ASAP. The higher the expansion rate the more impact on system performance.
Because a concatenated expansion does not need to re-stripe data the expansion rate has no impact on the
concatenated expansion or host I/O performance.
Stripe Element Size: If multiple LUNs are contained within a component within the metaLUN a new stripe
element size will be calculated. The size is based on the first LUN in the component and is calculated by
determining the stripe size of the first LUN (# of data disks x LUNs stripe element size) and multiplying that
by the Element Size Multiplier. For example, to determine the stripe size of a 4 + 1 RAID 5 LUN with a stripe
element size of 128 sectors, you multiply 128 by 4 for a stripe size of 512 sectors or 256KB of data.
Multiplying 256KB by an Element Size Multiplier of 40 (the default) will create a metaLUN element size of
10MB. FLARE would write 10MB of data to a LUN in the striped metaLUN component before proceeding to
next LUN in that component. The stripe element size for each component in the metaLUN can be different
since the stripe size of the first LUN in each component can vary. Valid element size multiplier values are 1
through 255.
Alignment Offset specifies the offset for a LUNs RAID stripe alignment when the host operating system
records private information at the start of the LUN. The default value is zero and this supports most host
operating systems. Alignment Offset is available only when the LUN you are expanding meets one of the
following conditions: The LUN is in or out of a Storage Group and has no existing data, or the LUN is not in a
Storage Group, has existing data, but you do not want to preserve this data.
Problems can arise when a host operating system records private information at the start of a LUN. This can
interfere with the RAID stripe alignment so that when data I/O crosses the RAID stripe boundary, the
storage-system performance is degraded. In order to properly use Alignment Offset, you must know the
exact size of the private information. Improper use will degrade performance.
13
2003 EMC Corporation. All rights reserved. 13 13
Expand LUN Summary and Confirmation
The final step in the wizard is to provide the user with a summary of the actions
about to performed. If any of the expand parameters need to be changed the
Back button can be used to access the appropriate wizard page. If all parameters
are acceptable, click Finish to initiate the expand action. The system will notify
the user that the expand has been initiated.
14
2003 EMC Corporation. All rights reserved. 14 14
Navisphere Manager View
At the conclusion of the expand activity, the new metaLUN is displayed in
Navisphere Manager with its component parts viewable. Note the icon change
and the renumbering of the component FLARE LUNs while the metaLUN
inherited the Base LUNs ID.
15
2003 EMC Corporation. All rights reserved. 15 15
Stripe Expansion Walk-Through
Performing a striped expansion is initiated in the same manner as a concatenated
expansion, right click the LUN to be used as the Base LUN and select Expand
from the menu.
16
2003 EMC Corporation. All rights reserved. 16 16
Expansion LUN Selection
This screen in the wizard represents the branching point between concatenated
expansion and striped expansion. For this walk-through Stripe Expansion is
selected. After the expansion type is selected, clicking Next will proceed. As with
concatenated expansion, this process can be aborted or backed-up at any point
until the final action is committed.
A major difference between concatenated expansion and striped expansion is the
LUNs presented for inclusion. Concatenated expansion was allowed to include all
free LUNs of any size while striped expansion may include only LUNs which are
exactly the same size as the Base LUN. For this walk-through two expansion
LUNs have been selected for inclusion.
17
2003 EMC Corporation. All rights reserved. 17 17
metaLUN Settings
As with the concatenated expansion, striped expansion allows the entire
metaLUN to become host visible, or size may be restricted to the current Base
LUN capacity and enlarged later, or a value between current and maximum may
be user selected. Remember, whole numbers only in this last option.
Here the metaLUN name has been changed from the default and the expansion
rate set at ASAP. The element size multiplier is left at the default value of 40 and
if the Base LUN has a default stripe size will result in 10MB of data being written
to a LUN within the metaLUN before proceeding to the next LUN in the metaLUN.
This will greatly reduce the amount of addressing calculations and disk seeks on
the backend and will result in greater performance than if the element size of the
metaLUN matched the Base LUN exactly.
18
2003 EMC Corporation. All rights reserved. 18 18
Expansion Wizard Final Step
The wizard provides a summary of the action to be performed and allows a final
chance to abort the action or change any and/or all the parameters. Once Finish
has been selected, the action will initiate and the system will provide confirmation.
19
2003 EMC Corporation. All rights reserved. 19 19
Striped MetaLUN Components
After the expansion is completed, the metaLUN will be visible in Navisphere
Manager. Here it can be seen that the metaLUN inherited the LUN ID of the Base
LUN which in turn was renumbered along with the expansion LUNs and that all
three LUNs comprising the metaLUN are considered to be a single component.
20
2003 EMC Corporation. All rights reserved. 20 20
MetaLUN Properties
MetaLUNs like FLARE LUNs have a Properties sheet. Right-click the metaLUN
and select Properties to open this dialog box. While the metaLUN is expanding
its progress may be tracked on the General tab. The Expansion Rate for an
expanding metaLUN may not be changed but can be changed after the current
expansion is complete. The User Capacity can also be altered on this page but
caution should be exercised when performing this action as some operating
systems may not contain functionality to perform dynamic expansion.
21
2003 EMC Corporation. All rights reserved. 21 21
Expanding RAID Groups
Drive Drive
1 1
Drive Drive
0 0
Drive Drive
2 2
Drive Drive
3 3
Drive Drive
4 4
Drive Drive
6 6
Drive Drive
5 5
Drive Drive
7 7
Drive Drive
8 8
Drive Drive
9 9
RAID 5 Group RAID 5 Group
Available Disks Available Disks
Drive Drive
1 1
Drive Drive
0 0
Drive Drive
2 2
Drive Drive
3 3
Drive Drive
4 4
RAID 5 Group RAID 5 Group
Drive Drive
5 5
Drive Drive
6 6
Drive Drive
7 7
Drive Drive
8 8
Drive Drive
9 9
Available Disks Available Disks
RAID Groups can be expanded even after creation and even when they
have active LUNs bound to them.
If a RAID Group has not reached its drive limit (highlighted earlier), it can be
expanded by adding Available drives. In other words, drives not belonging
to any RAID Group.
The above RAID 5 Group begins with 5 disks, which is below the RAID 5
limit of 16. Drives 5 and 6 are not currently in a RAID Group so they can be
added to this group. The end result is a 7-disk RAID 5 Group.
22
2003 EMC Corporation. All rights reserved. 22 22
Expanding RAID Groups Multiple LUNs
0 0 1 1 2 2 3 3 4 4
0 0 1 1 2 2 3 3 4 4 5 5 6 6
LUN 0 LUN 0 20 GB 20 GB
LUN 1 LUN 1 20 GB 20 GB
LUN 2 LUN 2 20 GB 20 GB
LUN 3 LUN 3 20 GB 20 GB
LUN 0 LUN 0 20 GB 20 GB
LUN 1 LUN 1 20 GB 20 GB
LUN 2 LUN 2 20 GB 20 GB
LUN 3 LUN 3 20 GB 20 GB
Free Space Free Space
Before Expansion
After Expansion
If there are multiple LUNs bound to a RAID Group, the LUNS will not change
size during expansion. They will be re-striped across all of the disks in the
expanded group and retain their original characteristics. The expansion
process creates a large pool of free space at the bottom of the RAID
Group for adding of additional LUN(s).
23
2003 EMC Corporation. All rights reserved. 23 23
Expanding RAID Groups Single LUN
0 0 1 1 2 2 3 3 4 4
0 0 1 1 2 2 3 3 4 4 5 5 6 6
LUN 0 LUN 0 80 GB 80 GB
LUN 0 LUN 0 140 GB 140 GB
Before Expansion
After Expansion
In this example, a single LUN is bound to the RAID Group, and it is bound to
the maximum capacity. When the RAID Group is expanded, and the option
to Expand LUN with Raid Group is checked, that LUN will be expanded.
Though the physical disk size will be increased for the host, some hosts will
not automatically stretch the partitions on the drive.
If the option to Expand LUN with Raid Group is not checked, the LUN will
remain the same size.
24
2003 EMC Corporation. All rights reserved. 24 24
Expanding RAID Groups
To expand a RAID Group, right-click on the RAID Group you wish to expand and select
Properties. Once in the RAID Group Properties box, select the Disks tab. That will show
all disks currently in the RAID Group. Click on the expand button to select from the
remaining unbound disks on that array. You will also see the option to Expand LUN with
Raid Group.
25
2003 EMC Corporation. All rights reserved. 25 25
Defragmenting RAID Groups
0 0 1 1 2 2 3 3 4 4 5 5 6 6
LUN 0 LUN 0 20 GB 20 GB
Free Space Free Space 20 GB 20 GB
LUN 2 LUN 2 20 GB 20 GB
LUN 3 LUN 3 20 GB 20 GB
Free Space Free Space 60GB 60GB
Largest LUN = 60GB
0 0 1 1 2 2 3 3 4 4 5 5 6 6
LUN 0 LUN 0 20 GB 20 GB
LUN 2 LUN 2 20 GB 20 GB
LUN 3 LUN 3 20 GB 20 GB
Free Space Free Space 80GB 80GB
Largest LUN = 80GB
After Defragmenting
A LUN can only be created using contiguous free space. In the top
example, a user has removed a LUN created earlier. This created a 20 GB
gap in the LUN stack. There also is 60 GB free space at the bottom of the
stack. The user can create a 20 GB LUN and a 60GB LUN, but he can not
create an 80 GB LUN.
If the user defragments the RAID Group, it will consolidate all Free Space in
the RAID Group, allowing for the creation of a single, larger LUN.
26
2003 EMC Corporation. All rights reserved. 26 26
Defragmenting RAID Groups
LUN 0 (5.000 GB). Offset 0.000 GB, Owned by - DATA-01
LUN 1 (5.000 GB). Offset 5.000 GB, Owned by - DATA-01
LUN 3 (5.000 GB). Offset 15.000 GB, Owned by - DATA-01
LUN 6 (5.000 GB). Offset 30.000 GB, Owned by - DATA-01
LUN 7 (5.000 GB). Offset 35.000 GB, Owned by - DATA-01
LUN 8 (5.000 GB). Offset 40.000 GB, Owned by - DATA-01
To Defrag a RAID Group, right-click on the RAID Group you wish to expand
and select Properties. Once in the RAID Group Properties box, select the
Partitions tab.
The Partitions tab lets you view the current partitioning of a RAID Group
and lets you defragment (consolidate the data on) the disks in a RAID
Group.
LUNs List of LUNs currently bound in this RAID Group.
RAID Group Partitions Graphical representation of arrangement of LUNs
and distribution of free space associated with this RAID Group.
Defragment Begins the RAID Group defragmentation process.
27
2003 EMC Corporation. All rights reserved. 27 27
Summary
metaLUNs
Striped
Concatenated
Expand / Defrag RAID Groups
Multiple LUNs
Single LUNs