Yes still the same

But will the VNX alerting if I configure only one Hot Spare for 36 disk?

Hot sparing for VNX is a little different on VNX from CLARiiON.  In addition the recommended ratio has been changed to explictly round-down to get the number of high-availablity hot spares.  Although the high availablity ratio remains 1:30.

Hot sparing is not enforced by the operating envirnoment. There are no alerts generated for having too low a ratio.   It is not required that you provision hot spares.  However, its prudent that you do so.

Below is the section on Hot Spares from the VNX Best Practices.

Hot spares

Hot spares are drives provisioned to automatically replace failing drives or drives that have abruptly failed.

Proactive sparing is the process of automatic replacement with a hot spare when a drive indicates it is likely to fail.  Failure indication comes from OE Block’s continuous diagnostics.  This is essentially a drive-to-drive copy operation.  It is very fast and uses a minimum of storage system resources.

Note that a hot spare is a temporary replacement for a failing or failed drive.  A RAID group operating with a hot spare is in degraded mode.  Degraded mode ends when the failed drive has physically been replaced, rebuilt, and the hot spare is returned to the pool of available hot spares.  A hot spare may or may not be identical to the drive it is replacing.  This may result in a difference between performance from normal operation.

Proactive sparing on the VNX automatically selects hot spares from a group of drives designated as hot spares.  Hot spares are global.  Global means they are shared by all the same type, in-use drives on the storage system.

Careful provisioning of hot spares ensures the efficient use of available capacity and best performance until a replacement drive is installed.  Otherwise, it is possible to have too many hot spares, or for a less appropriate drive to be used as a hot spare replacement.

For example, a 10K rpm SAS drive could automatically be selected as a spare for a failing 15RPM SAS drive.  This would result in a noticeable RAID group performance degradation.

Hot spares are specially provisioned drives.  There is no operating environment enforcement to allocate any drives as hot spares.  However, it is strongly recommended that hot spares be provisioned.

Hot spare algorithm

The appropriate hot spare is chosen from the provisioned hot spares algorithmically.  If there were no hot spares provisioned of appropriate type and size when a drive fails, no rebuild occurs.  (See the Drive Rebuilds section.)  The RAID group with the failed drive remains in a degraded state until the failed drive is replaced; then the failed drive’s RAID group rebuilds.

The hot spare selection process uses the following criteria in order:

1. Failing drive in-use capacity - The smallest capacity hot spare drive that can accommodate the in-use capacity of the failing drive’s LUNs will be used.
   
2. Hot spare location - Hot spares on the same back-end port as the failing drive are preferred over other like-size hot spares.
   
3. Same Drive type – Hot spare must be of the same drive type.
   

Failing drive in-use capacity

It is the in-use capacity of the failing drive’s that determines the capacity of the hot spare candidates.  Note this is a LUN-dependent criterion, not a raw drive capacity dependency.  This is measured by totalling the capacity of the drive’s bound LUNs.  The in-use capacity of a failing drive’s LUNs is not predictable.  This rule can lead to an unlikely hot spare selection.  For example, it is possible for a smaller capacity hot spare to be automatically selected over a hot spare drive identical to, and adjacent to the failing drive in the same DAE.  This occurs because the formatted capacity of the smaller, hot spare (the highest-order selection criteria) matches the in-use capacity of the failing drive’s LUNs more closely than the identical hot spare drive.

Note that a hot spare’s form factor and speed are not a hot spare criteria within the type.

For example, a 3.5” format 15K rpm drive can be a hot spare for a failing 2.5” 10K rpm SAS drive.

Hot spare location

Ideally, the automatically chosen hot spare will be of the same speed, and capacity as the failing drive.  This would prevent any loss of performance while the RAID group is in degraded mode while operating with a hot spare in the RAID group.  To influence this type of selection, ensure an appropriate hot spare is located on the same backend port of the drives you’d like it to replace.  This may require provisioning that segregates drives by type to different backend ports to make the most economical use of the fewest hot spares.  Note that the backend port attachment should always be verified when provisioning hot spares.

Same Drive type

may be an alternative.  A large number of drives are 32 or more in-use RAID groups.

Hot sparing best practices summary

EMC CLARiiON Global Hot Spares and Proactive Hot Sparing white paper available on Powerlink.

High availability, high performance hot sparing example

The storage system is to be provisioned with the following hard drives:

¨       5x 200 GB flash drives

¨       17x 7.2k rpm, 2 TB NL-SAS hard drives

¨       25x 10K rpm, 300 GB SAS hard drives

How many hot spares should be used to complete the provisioning for this storage system?  Where should they be positioned?

Configuring the storage system for hot spares

DPE/DAE

Back-end
port

System drives

Data drives

Hot spares

Total DAE
drives

DPE

0

4x SAS

20x SAS

1x SAS

25

DAE 1

1

0

15x NL-SAS

0

15

DAE 2

1

0

1x NL-SAS
4x Flash Drive

1x NL-SAS
1x Flash Drive

7

* Vault drives are not hot spared,
they have their own protection.

Total drives:

47

Total SAS drives

25

SAS hot spares ratio

1:20*

Total NL-SAS drives

17

NL-SAS hot spare ratio

1:16

Total Flash Drives

5

Flash Drive hot spare ratio

1:4

Effective hot spare ratio:

1:15*

Table 11 Hot spare provisioning example

A DPE-type enclosure is always on backend port 0.  Twenty-five 300 GB 10K RPM SAS drives are provisioned on it.  Four of these hard drives are used as the system drives.  Note that hot spares are not provisioned for the system drives.  This provides for 4x (4+1) SAS RAID groups.  One of the SAS drives is a hot spare.  Being on backend port 0 with the SAS drives, it has port affinity for sparing with them.

DAE1 is provisioned on backend port 1.  It is provisioned with 15x 7.2k rpm 2 TB SAS drives.  This provides for 1x complete (6+2) RAID group and part of another.

Configure DAE2 to be on backend port 1.  It is provisioned with the remaining 4x NL-SAS drives.  One NL-SAS drive completes the partial NL-SAS RAID group in DAE1.  The storage system can have 2x (6+2) NL-SAS RAID groups.  The second NL-SAS drive is a hot spare.  In addition, this DAE is provisioned with all the flash drives.  This includes four flash drives intended for use as a FAST Cache.  A fifth flash drive is for use as a hot spare.  Note that the NL-SAS hot spare cannot be as a spare for a failing SAS drive.  A SAS drive cannot be a spare for a failing NL-SAS drive.  Also that flash drives can only hot spare for flash drives.

In this provisioning the ratio of drives to hot spares is lower than 1:30.  This was done intentionally.  It provides for a higher level of performance in degraded mode should any of the several drive types in use fail.

If by saying system Drives are you reffering to the Flare Drives in which the OS is stored. If yes then the answer here is a Yes, The flare drives does not require a Hotspare for sparing as its Triple mirrored.

Thanks

Rupal Rajwar

Hi all!

Best Practices said that system drives of VNX doesn`t require hot spare. Is it the same for Clariion? Does clariion system drives require hot spare?

Just to expand a bit on this.  If user-defined LUN(s) are created on any/all the system drives in the remaining space, then a hot spare would need to be considered for those drives as that data will rebuild as necessary.

On the other hand, if the remaining space is not utilized, then as noted already, a hot spare is not required for those drives as the system partition does not get rebuilt onto a hot spare.

I see something that I didn't know: NL-SAS cannot be a hot spare for a SAS? And the other way around as well?

Hmmmmm, in the FC/SATA days these two could replace eachother. Good to know!