flare 32, RLP on POOL

Use different disk spindles for RLP. You are refering to Snapview Snapshots and not VNX Snapshots, correct?

The RLP's can Definetely be formed out of a pool but only as Thick LUNs.  A homogeneous Pool or RG of SAS drives is also recommended for the same.

But again for the best performance the Reserve LUNs should stay on traditional raid groups.

As the concept here still remains the same as COFW and if performance is the need of the hour for you then the go with the Traditional RG's.

Once the luns are assigned for RLP(or for any application), these luns become the private luns.

Thin LUNs cannot be private LUNS and therefore cannot be used for Reserved LUN Pools, Clone Private LUNs, or Write Intent Log LUNs.

For further reference on the same this can be reffered : https://community.emc.com/docs/DOC-14622

Thanks

Rupal Rajwar

used for replication, but we're hammering our 4+1R5 spindles during a a-sync replication:

What if we would create thick RLP luns on our pool(15 SAS,5 NL-SAS), load would be spread over more than these 5 drives in a single RG.

So what gives a better performance, a dedicated 5 drive RG or spread my RLP's over 20 drives in a pool.

If you do that then you are storing your current writes as well as the COFW on the same spindles. Not the best thing to do. EMC best practice is to use a different RAID Group/Pool for RLP in order to seperate the COFW.

If you decide to create the RLP LUNs in the pool make sure you bind the LUNs on the highest tier. Do not auto tier RLP LUNs since you don't want COFW to reside on the NL-SAS drives.

rupal,this document is not opening for me.

Is this accessible for EMC employees only ?

Also please refer to Primus article “emc202636” for more in-depth information on the Procedure to configure RLP in VNX or CLARiiON array.

~ Best procedure to Configure Reserve LUN Pool (RLP) on VNX or CLARiiON array | VERSION 1 | Created on: Mar 7, 2012 7:31 AM by Satish_Kutty - Last Modified:  Mar 7, 2012 7:31 AM by Satish_Kutty

Best procedure to Configure Reserve LUN Pool (RLP) on VNX or CLARiiON array

VERSION 1

Created on: Mar 7, 2012 7:31 AM by Satish_Kutty - Last Modified:  Mar 7, 2012 7:31 AM by Satish_Kutty

Product:

CLARiiON AX Series, CLARiiON CX Series,CLARiiON iSCSI Array, VNX, VNXe

Description:

ATA, SATA2 and LCFC drives are not generally recommended for reserved LUN pools.  While they can work in some cases, their lower performance make them a bottleneck, especially when source LUNs are bound on FC RAID groups and the host is writing at a high rate (resulting in a large amount of COFW activity).Using the Vault drives (0.0.0 to 0.0.4) for the RLP is also not recommended.

Using Pool LUNs for the RLP is supported, but these LUNs should not be Thin or have Compression or FAST Cache enabled.Care should also be taken to make sure that the Pool used does not contain any Source LUNs (i.e. LUNs that have SnapView, MirrorView/A or SAN Copy sessions running on them).

Resolution:

Using Pool LUNs for the RLP is supported, but these LUNs should not be Thin or have Compression or FAST Cache enabled.Care should also be taken to make sure that the Pool used DOES NOT contain any Source LUNs (i.e. LUNs that have SnapView, MirrorView/A or SAN Copy sessions running on them).

Snap Cache LUNs are allocated to a Snap Session according to their Array Logical Unit Number.  Therefore if three RAID groups were to be used for the RLP and the LUNs were bound sequentially in each RG, then most of the performance load would fall upon the first of these RG, because its LUNs would have the lowest Numbers.  To work around this, the LUNs can be bound one at a time across the RLP RG in a “round robin” fashion” and gaps can be left in the numbering to allow for more RG to be added to the RLP in future.

In the example below, we have allocated RAID groups to the RLP, but we have left room for up to seven more to be added to our RLP numbering scheme later:

Array LUN     RAID Group

800                  80
801                  81
802                  82
810                  80
811                  81
812                  82
820                  80 and so on.

The next RAID Group to be added would number its RLP 803, 813, 823 etc.

The size of the LUN is not taken into account and for that reason, it is best to make sure that all RLP LUNs are the same size.

Also please refer to Primus article “emc202636” for more in-depth information on the Procedure to configure RLP in VNX or CLARiiON array.

27 Views Categories: CX-Series Tags: clariion

Vipin : Its posted in the Forum so its not Internal to EMC.

Thanks

i can't open i either.

Rupal, It gives the error

" It appears you're not allowed to view what you requested. You might contact your administrator if you think this is a mistake "

and I have heard EMC internal links are also available in ECN.

Cant open the document either

Doesnt work for me either.

Thanks Baif for posting this out here.

This was the same link i refferred which i think no is able to access.

Thanks

Hello John,

Reserved LUN Pool is no longer required for VNX Snapshots, it use space for snapshots in the same storage pool as their base LUN.

VNX Snapshots are limited to pool based provisioned LUNs (i.e. not RAID Group based LUNs)

Thanks

Hi Shawky,

Does this mean it is possible to use VNX snapshots as a source for Mirrorview a-sync?

And in other words: when sizing a new system just take a snapshot reserve in account within the pool without the need for dedicated spindles/RLUN pool.

(have been struggling with this question for some time now, would be great to be certain )

Thanks!

Hi Daan@Proact,

Regarding your question for VNX snapshots, they cannot be used as a source for MV/A.

Regarding the new technology used in VNX snapshots, It is recommended to consider the space availability in pools, if you are planning to use VNX snapshots.

You can refer to MV Knowledgebook white-paper

https://support.emc.com/docu32906_White-Paper:-MirrorView-Knowledgebook:-Releases-30---32---A-Detailed-Review.pdf?language=en_US

With Regards