This post is more than 5 years old
28 Posts
0
777
March 7th, 2016 11:00
RecoverPoint with XtremIO- Questions
Hello,
I have a few questions based on the documentation and release notes.
- Adding a replication set/ Removing a replication set/ Resizing a volume in a replication set within a CG causes CG to be refreshed and Snapshots to be deleted. Are the snapshots in XtremIO deleted instantly? Are there any other workarounds to prevent this from happening?
- When removing a Replication Set from an existing CG, are the corresponding volumes removed from XtremIO CG which RP crates automatically?
- Let's say the replication mode is periodic and the interval is set to 1min. When will the next snap be created? One minute after the creation of first snap or one minute after the creation and complete shipping of the first snap?
- Why is that the actual RPO is twice that of the mentioned duration when the replication mode is set to "Periodic"?
- Is there any advantage in enabling compression/ dedup/ io throttling in RecoverPoint when implementing XtremIO snap based replication?
- Are the WAN throughput limits mentioned in the Preformance Guide still applicable when implementing XtremIO snap based replication?
No Events found!
Idan
675 Posts
0
March 7th, 2016 14:00
Hi there,
1. The snaps in these scenarios would be deleted as a background process. Note that in RP 4.4 and onwards, adding a RSET does not inflict journal history loss but only a volume sweep of the newly added volume.
2. The volumes which RP had created should be removed in such a scenario.
3. The interval is counted from the time transfer has started.
4. The worse case RPO in Periodic is twice the interval since it's possible that w/ transfer interruption, the RPO would be the amount of time passed by the latest PiT on the target.
5. Yes to all - Dedup and compression - b/c RP is reading the changed data and compresses and dudups across the WAN (the fact that it's deduplicated and compressed on XtremIO isn't directly related to the transfer made by RP). IO throttling should be used to minimize impact on production during initialization by limiting the read throughput from production and/or replica volumes
6. Not sure what perf guide you're referring to, if it's the 4.4 Performance one then yes but incoming write rate limitations and other scale/perf limits are different for XtremIO.
Regards,
Idan Kentor
RecoverPoint Corporate Systems Engineering
Cherry251
28 Posts
0
March 7th, 2016 14:00
Thank you Idan. I was referring to the EMC RecoverPoint 4.1 Performance Guide. This guide doesn't cover the snap based replication with XtremIO. Is there any document which talks about the performance limits specific to RP 4.1.2 with XtremIO?