If you have switched-dynamic-RDF already enabled (check output of "symcfg -sid xxx -v list") you can simply use symrdf and remove dirs 4D and 13D from groups 1 and 2 (use "symrdf modifygrp"), then use (again) symrdf to create a brand new RDFG ("symrdf addgrp") using only dirs 4D and 13D. Last step is delete pairs ("symrdf susp" and "symrdf deletepair") and then create them again on the new group ("symrdf createpair").
When you have your brand new pairs on new group, put them in adaptive copy ("symrdf set mode acp_disk"), establish the pairs and when you have little if no tracks waiting on R1 side, put the group (the DG that you use to manage the pairs) in async mode ("symrdf set mode async").
Since code 5671 (and it looks like you are at least at 5771) each RDF processor can run both SRDF/S and SRDF/A. Thus I see no real issues with your RDF configuration. But do considere that SRDF/A puts lot more pressure on cache. Thus ask EMC and check carefully if both R1 and R2 boxes have plenty of cache. Another thing to considere is that R2 performances may impact SRDF/A, thus ask EMC to check if R2 is fast enough.
In a word: Yes you can. But you probably need to spend some money
Ferry just to be frank ... Actual configuration looks a little "confused" to me
Processors 3C and 14C still have a couple of configured groups (6 and 7) but remote box isn't available.
Processors 4D, 7D, 8D, 9D, 10D, 13D and 14D have an 1:1 relation with remote processors and all run only group 1.
Processor 3D talks with multiple remote processors (at least 5, up to 8) and runs both group 1 and 2.
Let's come back to your question.
SRDF/A is simply a different mode that you can use on any existing group. The main "issue" is that with SRDF/A all devices belonging to a group must run SRDF/A. I mean you can't mix SRDF/S and SRDF/A devices in a single RDF group. Thus you probably have to move devices to different groups (and of course create new groups)
It's far better better thinking about using different processors for SRDF/A and SRDF/S groups (since they have different "workloads" and may interfere one with the other).
You are right sorry for the confusion, Processors 3C and 14C where used for remote replication and are not used at this moment. Processors 4D, 7D, 8D, 9D, 10D, 13D and 14D have the same setup as Processor 3D ( i used .... because else it would take to much space ) and you are right those Processors talk to 8 remote processors over RA 1 and RA 2 group.
So if i understand you correct the steps to follow would be to use for example Processors 4D and Processors 13D for SRDF/A and create a new RA group for that, deletepair for curent SRDF/S database disks and createpair over new RA-group for those devices again.
Processors 3C and 14C where used for remote replication and are not used at this moment.
Ok .. cleaning up the things may require a SR and a PSE... :-/
Processors 4D, 7D, 8D, 9D, 10D, 13D and 14D have the same setup as Processor 3D ( i used .... because else it would take to much space ) and you are right those Processors talk to 8 remote processors over RA 1 and RA 2 group.
Thank you, now everything is crystal clear
Processors 4D and Processors 13D for SRDF/A and create a new RA group for that, deletepair for curent SRDF/S database disks and createpair over new RA-group for those devices again.
Yup you are right. And don't forget to change groups 1 and 2 removing processors 4D and 13D.
Since code 5773 you can even forget about deleting/recreating pairs and simply use the brand new interesting (and working) "movepair" feature...
Yes code 73 would be great, but if have some fighting to do for that because a lot of servers running now need to be upgraded for that. So i dont see that happen :- (
Can you show me in the direction for the commands you use to start using 4D and 13D for SRDF/A?
xe2sdc
6 Operator
•
2831 Posts
1512
0
Posted October 17th, 2009 13:00
When you have your brand new pairs on new group, put them in adaptive copy ("symrdf set mode acp_disk"), establish the pairs and when you have little if no tracks waiting on R1 side, put the group (the DG that you use to manage the pairs) in async mode ("symrdf set mode async").
Enjoy