armin2504
1 Nickel

clarification on RP use

storage_group.PNG.pngCustomer wants to do a recovery test for 2 of their servers ( AIX and Linux ) . The replication is from xtremIO to VNX using RP (4.4)i have been provided with the Host to Lun mapping list . The test servers on the DR  will be zoned at the time of the test .

I am little  familiar with the testing a copy procedure . What I do not understand is that do I have to create a separate storage group for the test and assign the copy luns again in that storage group for the DR servers to access . I already see a SG created with name _RPA_SG that has Journal luns and copy luns . All The luns are boot from san Luns

Could some one tell me the procedure what needs to be done for the test in my case ? Attached is the screenshot .

I have an additional question. When we do a test on CG , how are the replication sets in the CG also participate in the test ?

Please explain the process if possible

Labels (1)
Tags (2)
0 Kudos
9 Replies
Highlighted
fadliz
2 Iron

Re: clarification on RP use

yes, you need to mask the LUN to the DR hos by creating a DR host SG and giving access only to the replica LUNs.

armin2504
1 Nickel

Re: clarification on RP use

So I have 2 hosts AIX ( HACMP)  and Linux ( VCS) cluster  I need 2 separate Storage groups for each of the clusters .Does these clusters  need a separate zoning as compared to how  we zone  regular hosts?The lun should be accessible to both the hosts in the cluster .

please clarify . Thanks !

0 Kudos
forshr
3 Zinc

Re: clarification on RP use

Yes, the hosts will need to be zoned to the VNX SP ports in order to for their initiators to be connected to the SGs.

You also need to add the replica volumes to the RP SG.

Regards,

Rich

0 Kudos
armin2504
1 Nickel

Re: clarification on RP use

Hello  Thanks for the suggestion .Customer is replicating from XtremIO to VNX .Is it  advisable to pause the replication and take a snapshot and then resume the replication . Is there any issue is doing this ?  Would like to know the best practices , thanks

0 Kudos
forshr
3 Zinc

Re: clarification on RP use

I don't follow? RP is taking a snapshot from the XtremIO array as part of the replication cycle.

Regards,

Rich

armin2504
1 Nickel

Re: clarification on RP use

Thanks Rich . RP does a snap based replication fr the xtremio storage

0 Kudos
forshr
3 Zinc

Re: clarification on RP use

Are you asking if you can generate an XtremIO array based snapshot whilst RP is generating a snapshot?

0 Kudos
armin2504
1 Nickel

Re: clarification on RP use

No . I was  just confirming to what you said about the Recoverpoint taking snapshot from the XtremIO array as part of the replication cycle.


I have one  small query related to this test . The DR test will be for 48 Hours . Is it recommended to do a direct access mode after logged access? I understand journal is not kept and a full sweep is done in direct access mode . Does that mean all the journals  for the copy  at the DR will  be lost ? if not when is the direct access mode recommended .


thanks!

0 Kudos
forshr
3 Zinc

Re: clarification on RP use

Direct access mode is recommended when there is a probability of the image access log filling which may happen with the conventional journal used on the VNX. This is dependent on a number of factors though. Firstly, the amount of writes versus the size of the image access log as a default 20% proportion of the journal. Secondly, if in fact the default of 20% is used. This cannot be changed dynamically by the way. it has to be done whilst the CG is disabled or with multi-copy, whilst the replica copy is disabled. However, even if it has been changed from the default to a higher number (up to 80%) it is again proportionate and dependent on the overall size of the journal.

You have listed the side-effects relating to direct image access and losing the journal PiT history for the replica copy journal is one of these side-effects. The alternative to this is to clone the replica volumes on the VNX once in image access mode, then disable image access mode and then use the clones for the testing.

Regards,

Rich