This post is more than 5 years old
3 Posts
0
1367
May 14th, 2014 09:00
EMC SRDF
SRDF - Normally its the delta from R1 --> R2, (bi-directional). Its easier to take a SNAP or Clone of R2 and do the DR testing and just drop the clone / SNAP its much safer.
However, I have a client that only has the EMC source (R1) and a remote target (R2) without an R3 or R4 for DR exercise. For a DR Exercise they stop replication (R1 --> R2) prior to a DR Exercise, and use R2 during the DR Exercise, and update R2 during the DR Exercise,,
They do not purge R2 updated data after the exercise, but want to restart replication post exercise.
My question is; will SRDF just send the delta (build up from prod while replication was suspended) from R1 to R2, or will SRDF hiccup and cause the R2 files re-synched from the beginning??
I know this is a long question but I hope I gave you enough to answer.. I have team members that state only the delta will be pushed during the re-sync but it does not make sense to me.
Thanks
seancummins
2 Intern
•
226 Posts
0
May 14th, 2014 10:00
It will only send the deltas over. When you're in a split state (replication is stopped and the R2s are Read/Write), both the R1 and R2 arrays will be keeping track of changed tracks on both sides. As tracks on either side are changed (by host writes), those tracks will be marked invalid on either the R1 or R2 volumes. When you resume replication, SRDF will merge the invalid track tables, and will incrementally copy all invalid tracks from the R1 to the R2.
Thanks,
- Sean
storarchitect
3 Posts
0
May 14th, 2014 10:00
Thank you for the information, very helpful!
seancummins
2 Intern
•
226 Posts
2
May 14th, 2014 10:00
Only if you explicitly reversed the direction of the replication so it was going 'backwards' from R2 to R1 (e.g. with a symrdf update) or performed a personality swap, reversing the roles of R1 & R2. If you just perform a symrdf establish, the R2 invalid track table will be merged into the R1 invalid track table, and replication will incrementally resume from R1 to R2.
storarchitect
3 Posts
0
May 14th, 2014 10:00
So dumb question... Is there any way that updated R2 data could bleed into R1 (Prod) when re-sync after DR exercise??