NetApp

1
Maximising Usable Space With Block Protocols
Peter Mason pmason_at_netapp.com
2
Agenda

• Whats the problem?
• Terms and Concepts
• Space Guarantees
• Sparse LUNs, Space Reclaimer
• Snap Reserve
• Fractional Reserve
• Vol Autogrow and Snap Autodelete
• Suggested configurations
• Recommendations

3
The Problem

• Usually noticed if using LUNs
• Only happens after first snapshot
• Due to current ONTAP defaults
• Can result in around 50 lost space
• But..
• Easy to rectify

Unfortunately, not well explained in the product
manuals
4
How other vendors handle data change
1000AM
Root inode
A
B
D
C
C1
5
How other vendors Snapshot works
1000AM
1100AM
1105AM
Root inode
Snapshot Read only
A
B
D
C
C1
6
How other vendors Snapshot works
1000AM
1100AM
1105AM
Root inode
Snapshot Read only
A
B
D
C
C1
7
How other vendors Snapshot works
1000AM
1030AM
1105AM
1100AM
Root inode
Snapshot1 Read only
Snapshot2 Read only
A
B
D
C
C1
C
8
How a NetApp data change works
1000AM
Root inode
A
B
D
C
C1
9
How a NetApp Snapshot works
1000AM
1100AM
1105AM
Root inode
Snapshot Read only
A
B
D
C
C1
10
How a NetApp Snapshot works
1000AM
1030AM
1105AM
1100AM
Root inode
Snapshot1 Read only
Snapshot2 Read only
A
B
D
C
C1
11
Terms and Concepts
Snapshots taken on a NetApp storage array are
stored in the same volume as the primary
data Snapshots taken on other vendors arrays
are stored in a separate snapshot area On NetApp
if you fill your volume with snapshot data your
LUN is taken offline to protect it Default
settings ensure this wont happen On other
vendors arrays if you run out of snapshot space
the snapshot is failed
12
Terms and Concepts FlexVol Guarantees

• Volume     
• Default guarantee for any new volume
• Space reserved in aggregate at volume creation
  time
• This space not available to other volumes
• Fractional Reserve settings can only be changed
  using Volume guarantees
• None        
• No space is reserved in aggregate at volume
  creation time
• Space is taken from the volume as data is written
• Space allocated first come, first served.
• Any individual volume still is not allowed to
  take more space than its specified size
• Permits thin provisioning at the volume level
• File         
• The same as none but any individual LUNs file (or
  LUN) is guaranteed space when created. Not widely
  used.

Stick with guaranteeVolume unless you have good
reason to do otherwise
13
Flexible Volume Space Management

• Space Guarantees

none
volume
file
Aggr
Aggr
NOTE The blue blocks represent space not disks
14
Space Guarantees Set to none
none
NOTE In reality, particular blocks are not
claimed. The volume is simply given a space
limit, and space is filled at random.
15
Space Guarantees at FlexVol Creation

• To specify a space guarantee at the creation of a
  FlexVol
• Example
• vol create sample_vol sample_aggr s file size
  10g

16
Space Guarantees Change or Update

• To change the space guarantee for an existing
  FlexVol
• Example
• vol options sample_vol guarantee none

17
Terms and ConceptsLUN Reservation (Sparse LUNs,
Thin Provisioning)

• Determines when space is taken out of the
  volume
• At LUN creation (i.e. fully provisioned)
• When data is written to the LUN (i.e. thin
  provisioned)
• Equivalent of guarantees for LUNs
• enable volume
• disable none
• Enabled by default
• Only to be disabled for thin provisioning of LUNs
• lun set reservation enable/disable
• lun create o noreserve (creates a thin
  provisioned LUN)
• Can change after LUN creation
• Command line only

18
Space and Sparse LUNs
32GB LUN holding 1GB data
Volume 16GB
19
Sparse LUNs Space Reclaimer

• SnapDrive for Windows includes Space Reclaimer
  (SR)
• SR turns fat sparse LUNs thin again
• SR only operates when a threshold value is
  reached
• SR will never work on hugely over-provisioned
  volumes
• SR should be used regularly on Volumes containing
  sparse LUNs
• SR is filesystem specific only NTFS currently
  supported
• Documentation for SR is minimal

20
Terms and Concepts - Snap Reserve

• Set at the volume level
• ONTAP removes the defined percentage from the
  volume
• Does not allow LUNs to be created or files to
  grow into the snap reserve space
• Snapshot data fills this space
• Traditionally set to 0 for volumes used with
  block protocols, but does not have to be zero in
  SAN environments.

21
Snap Reserve
Figure 1 - DEFAULT
Figure 2
Historically recommend setting Snap Reserve to 0
for block storage, but there are now reasons to
reconsider this policy
22
Terms and Concepts - Fractional Reserve

• Set at the volume level (default is 100), can
  only be altered if space guarantee volume, can
  be set between 0 and 100
• Allows thin provisioning of snapshot space
• Independent of Snap Reserve
• Determines amount of space ONTAP reserves for
  snapshot data for LUNs
• Space is reserved when the first snapshot is
  taken. Space reserved used space in LUN(s)
• Fractional Reserve Space is only used if no other
  space in the volume is available

23
Fractional Reserve - Default Example

• Create the space reserved LUN
• lun create s 100gb t aix /vol/vol1/lun1

Fractional Reserve 100
guarantee volume space reservation enable
25 GB 100 GB LUN1 LUN1 vol1 reserved full
24
Fractional Reserve - Default Example (cont.)

• Fill the LUN

guarantee volume space reservation enable
25 GB 100 GB LUN1 LUN1 vol1 reserved full
25
Fractional Reserve - Default Example (cont.)

• Take the Snapshot
• snap create lun1 snap1
• Space is Reserved

guarantee volume space reservation enable
25 GB 100 GB LUN1 LUN1 vol1 reserved full
26
Fractional Reserve 100 - LUN Not Filled

• What if the LUN is not full when the snapshot is
  taken?
• Only the amount of space filled is reserved

guarantee volume space reservation enable
25 GB 100 GB LUN1 LUN1 vol1 reserved full
But the amount of data in the LUN might not be
the same as what the host reports..
27
Fractional Reserve 25 Percent Example

• Set Fractional Reserve
• vol options vol1 fractional_reserve 25

Fractional Reserve 25
guarantee volume space reservation enable
25 GB 100 GB LUN1 LUN1 vol1 reserved full
28
Fractional Reserve 25 Percent Example (cont.)

• Take Snapshot
• Space is reserved

guarantee volume space reservation enable
25 GB 100 GB LUN1 LUN1 vol1 reserved full
29
Fractional Reserve 25 - LUN Not Filled

• What happens if fractional reserve is 25
  percent and the LUN is 50 percent full when the
  snapshot is taken? 12.5 GB is reserved

guarantee volume space reservation enable
25 GB 100 GB LUN1 LUN1 vol1 reserved full
30
Terms and Concepts - Snapshot Auto Delete

• Set on a per volume basis
• Determines when (if) snapshots will be
  automatically deleted
• Makes NetApp like competition but better
• snap autodelete ltvol-namegt on off show
  reset help
• snap autodelete ltvol-namegt ltoptiongt ltvaluegt...
• Supported options and corresponding values
• commitment try, disrupttrigger volume,
  snap_reserve, space_reservetarget_free_space 1-10
  0delete_order oldest_first, newest_firstdefer_de
  lete scheduled, user_created, prefix,
  noneprefix ltstringgt
• Snapshots locked by clones, cifs or restores will
  not be deleted change planned for 7.3

31
Terms and Concepts - Volume Autosize

• Set on a per volume basis
• Determines if a volume should grow when nearly
  full
• Possible settings
• ON
• Maximum size (default 120 of initial size)
• Increment size (default 5 of original size)
• OFF
• vol autosize ltvol-namegt -m ltsizegtkmgt
• -i ltsizegtkmgt
• on off reset
•  Try_first option determines whether auto_grow or
  auto_delete is attempted first
• The possible try_first values are
• snap_delete
• volume_grow
• e.g. vol options volname try_first snap_delete

32
AutoSize and SnapMirror

• Beware growing volumes which are in a SnapMirror
  relationship
• The trick is to use volume space guarantee of
  none at the DR site
• Then set the size of the volumes at DR to the
  size of the aggregate
• When failover is invoked, volumes at DR inherit
  size and characteristics of the primary site
  volumes

33
Space Management Default Configuration

• Default Configuration
• Guarantee volume
• LUN reservation on
• Fractional_reserve 100
• Snap_reserve 0
• Auto_delete off
• Auto_grow off
• Try_first NA

Available space
40GB
Fractional Reserve
30GB
30GB
TestLUN
34
Space Management Default Configuration

• Positives
• Easy to manage/monitor space
• Running out of space only results in no
  additional snapshots being created
• SnapShots and active LUNs will always be
  available
• Volumes are independent of each other
• Negatives
• Requires 2X ? space
• Expensive

35
Thin Provisioning Snapshot SpaceVol Auto Size

• Auto Grow Configuration
• Guarantee volume
• LUN reservation on
• Fractional_reserve 0
• Snap_reserve 0 (or X)
• Auto_delete optional
• Auto_grow on
• Try_first auto_grow

30GB
Available space
LUN
70GB
Preferred Configuration
36
Thin Provisioning Snapshot Space Auto Size

• Positives
• Uses shared free space from the aggregate for
  possible growth
• Very space-efficient
• Can tune amount of snapshot space per volume
• Works even with a smaller number of volumes
• Negatives
• Volumes are not completely independent of one
  another
• Always the case when using shared free space
• Growth in one volume can limit growth of another

37
Volume autosize Example

• vol autosize vol1 m 800gb i 150gb on
• The current volume is 500GB
• There are 2 LUNs (160GB)
• So with these 4 Snapshot copies, the volume is
  98 full

Snapshot 4 65GB
Snapshot 3 60GB
Snapshot 2 75GB
Volume is 98 full
Snapshot1 100GB
LUN2 95GB
LUN1 - 95GB
38
Volume autosize Example (cont.)

• vol autosize vol1 m 800gb i 150gb on
• Now vol1 is 650GB, leaving 25 percent free space

Snapshot 4 65GB
Snapshot 3 60GB
Snapshot 2 75GB
Snapshot1 100GB
LUN2 95GB
LUN1 - 95GB
39
Volume autosize Example (cont.)

• vol autosize vol1 m 800gb i 150gb on
• Snapshot5, 145GB is added to the volume

Snapshot 5 145GB
Snapshot 4 65GB
Snapshot 3 60GB
Snapshot 2 75GB
Snapshot1 100GB
LUN2 95GB
LUN1 - 95GB
40
Volume autosize Example (cont.)

• vol autosize vol1 m 800gb i 150gb on
• Vol1 800GB Max Growth

Snapshot 5 145GB
Snapshot 4 65GB
Snapshot 3 60GB
Snapshot 2 75GB
Snapshot1 100GB
LUN2 95GB
LUN1 - 95GB
41
Thin Provisioning Snapshot Space Autodelete
SnapshotAvailable (Snap Reserve)
20GB

• Auto Delete Configuration
• Guarantee volume
• LUN reservation on
• Fractional_reserve 0
• Snap_reserve 20
• Auto_delete snap_reserve
• Auto_grow recommended
• Try_first snap_delete

Available space
20GB
LUN
60GB
Alternate Configuration
42
Thin Provisioning Snapshot Space Autodelete

• Positives
• Easy to monitor/understand space just volume
  (and .snapshot)
• Sacrifices snapshots before active LUNs
• Volumes are independent of each other
• Looks like the competition
• Negatives
• Doesnt use shared space from the aggregate
  (unless auto_grow is also used)
• Until ONTAP 7.3 have to be cautious with using
  FlexClones
• Beware, space is not returned immediately
• (significant improvements in ONTAP 7.3)

43
Filling Volume without autodelete
Starting state - 100GB Volume - 2 x 40GB
LUNs - snap_reserve 20 -
fractional_reserve 0
Writing data to LUNs Create snapshot 1 Create
snapshot 2 Volume is 100 full Writes to LUNs
are returned with out of space error
messagesApplication crashes!!
LUN 2 40GB
LUN 1 40GB
44
Filling Volume with autodelete
Starting state - 100GB Volume - 2 x 40GB
LUNs - snap_reserve 20 -
fractional_reserve 0 - trigger
snap_reserve - delete_order oldest_first
LUN 2 40GB
Writing data to LUNs Create snapshot 1 Create
snapshot 2 Snapshot 1 is autodeleted Snapshot
2 is autodeletedSnapshots are lost but
application continues running
LUN 1 40GB
45
(No Transcript)
46
Method 1 Auto_Grow First (Recommended!!)

• Make sure you include adequate PS resourcing for
  initial set-up
• Target snapshot space at 20-70
• Set fractional reserve to 0, volume guarantee
  volume..
• Set Snap Reserve to 0
• Enable vol auto size and optionally snap auto
  delete (triggervolume)
• Set try_first volume_grow
• Change default setting of wafl_reclaim_threshold
  to 90-95 (default 98) remember the trigger for
  autogrow is always volume. NO NEED TO DO THIS
  WITH ONTAP 7.2.4 OR LATER
• Keep between 10 and 40 spare space in your
  volume (this is used for snapshot data but is not
  reserved for that purpose)
• Put the remainder of your free space in your
  aggregate (this becomes global free space)
• Snap autodelete is used only when the volume
  reaches max size or there is no more space in
  aggregate. Do not use autodelete at all if
  SnapMgr data is in the volume

47
Method 2 AutoDelete First

• Make sure you include adequate PS resourcing for
  set-up
• Target snapshot space at 20-70
• Set fractional reserve to 0, volume guarantee
  volume.
• Set Snap Reserve at default 20
• Enable snap auto delete (triggersnap_reserve or
  triggerspace_reserve) and vol auto size
• Set volume option try_first snap_delete
• Change default setting of wafl_reclaim_threshold
  to 90 (default 98) if triggersnap_reserve, or
  lower if space_reserve
• You may elect to have some free space in your
  volume. Put the remainder of your free space in
  your aggregate (this becomes global free space).
• Vol auto size is used as a last resort in case
  the snapshots I have cannot be deleted
• Over time I may manually reallocate some of my
  global space into selected volumes to increase
  LUN sizes or provide more snap reserve
• BEWARE issues with SnapMgr products and
  Autodelete DO NOT use autodelete on any volumes
  containing SnapMgr data

48
SnapManager Exchange/SQL

• DO NOT use snap auto delete
• Instead use SME and SMSQL equivalents which only
  operate on fractional reserve usage at this time
• Set Fractional Reserve to a low value (e.g. 5)
• This value is dependent on the volume size
• Set higher for small volumes (under 20GB)
• Set the trigger value in SnapManager to 1
• Immediately any fractional reserve is used,
  snapshots are deleted.

49
Summary

• Using the techniques in this presentation can
• increase your storage utilisation significantly!