Start a Conversation

Unsolved

This post is more than 5 years old

1175

January 26th, 2016 01:00

Limit of luns for journals its 256?

Hi,

we have an environment, production site VNX 5600 --RPA (vnx is the spliter) / replication site other VNX 5600 -- RPA (vnx is the spliter)


In replication site we have a Raid Group with sas disk, and We arrived as limits of luns for journal what is 256.


Wich is the recomende size of lusn or organization for configure journal disk?.

the enviroment don't have more changes, is very static.


Thanks.




2 Posts

January 26th, 2016 05:00

hi,

the limit i have in VNX 5600 in Raid Group dedicated with sas disk. I don't create more luns of 256.


limit.png

limit_Raid_group.png

2 Intern

 • 

522 Posts

January 26th, 2016 05:00

Hi,

I think the number of journal volumes per copy in a CG is 128 from the release notes, not 256 - did you add 256 LUNs per a single copy?

In terms of the optimal size, this can depend on the change rate of the CG (incremental changes - in your case sounds like this is minimal) and the desired protection window required for this CG (the amount of continuous PiT snapshots you want to keep on the target side). If you don't have the information to size these (or have EMC size them for you with the performance data), the rule of thumb is generally 15-20% of the total size you are replicating to be allocated for journal space. For example, if you are replicating 1TB of data from the source, the RoT for jvols would be 150-200GB per copy. Of course this can be less and you can use trial and error to allocate more or less space depending on what you prefer for a desired protection window. I believe here is a calculator to assist with this, but it requires knowing some information about your application profile to fill out.

In terms of the size of the actual LUNs you allocate for this jvol space, this too will depend on how big your jvol space is. For example, if you deem 200GB to be the total jvol space needed on both sides from the example above, then creating 4 x 50GB LUNs for this jvol space is likely ideal. You will want to try and maintain the following in terms of best practices:

  • Journal volumes should be non-thin LUNs.
  • If more than one volume at a time is added to the journal, it is recommended that all added volumes be the same capacity for best performance and efficiency. If the added volumes are the same or nearly the same capacity (at least 85% of the largest volume), data is striped across those journal volumes, improving performance.
  • LUNs from different RAID groups/spindles/disk arrays can be used to get better performance, so using multiple journal volumes would be beneficial.
  • Ideally, allocate the same journal space on both copies to ensure that during a DR event, you will have the same SLA on source side with the allocated journal volumes.
  • Multiple smaller LUNs are preferable over fewer larger LUNs or a single LUN - this is due to striping, but also to future growth where if you need to account for an increase in workload on the source or an extension of the protection windows on the target, you don't have to over allocate space and can simply add it in smaller increments. Using the example above, you would allocate another 50GB volume, versus maybe a 100GB or 200GB volume when the time comes.

Check out the RP Administration guide as there is some of this material contained in there for setting up journals and adding volumes to existing journals. Your EMC SE could also assist here as well.

HTH,

-Keith

2 Intern

 • 

1.1K Posts

January 26th, 2016 05:00

Just for clarification, are you stating that you have hit a limit of 256 journal volumes for a CG copy?

No Events found!

Top