This post is more than 5 years old
18 Posts
0
6172
RecoverPoint - Test a Copy - Logged Access
Hi RecoverPoint community,
I am seeking confirmation on this as the documentation is a bit fuzzy around this. I am working on a test DR plan at the moment and we have enabled logged image access for all CGs involved in this test. Here is my understanding and what I need confirmation on:
- Logged Image Access depends on the change rate and not so much the 10% of protection window rule of thumb (48 hours window = 4.8 hours of logged access)
- If Logged Access fills to 99%, the host will lose access to the tested copy image
- If Journal Capacity fills to 99%, replication is paused and a full sweep will be needed once we are done
We changed a few to direct access mode once we noticed is was approaching ~70% logged access capacity. I'm just curious if I should change them all to direct access mode or keep monitoring like we are and change them on a case-by-case basis?
forshr
1.1K Posts
1
March 1st, 2016 01:00
Hi Chris,
You are alerted at specific ceiling intervals re the logged access progress. When journal capacity is reached and the RPA memory is full, replication is paused and marking is initiated at the relevant Prod copy journal. This is not a full sweep.
If you switch to direct access mode, replication is automatically paused and the journal history is discarded. When you disable from direct access mode there is a 'short' initialization so that Prod and the Replica can ne synched both in terms of new writes accumulated at the Prod side and recognition of changed blocks on the Replica side during the period of direct access. These have be re-synched from the Prod side.
Regards,
Rich
chrismahon
18 Posts
0
March 1st, 2016 05:00
Thanks for this! This is extremely helpful. Unfortunately for this customer, a tertiary site is not available and there is no extra capacity for a replica copy. The only option was direct image access. I'm sure discussions will be had with them about these options as this DR test is a yearly occurrence.
ankur.patel
286 Posts
1
March 1st, 2016 05:00
No problem, just to be clear there are only 2 sites in all scenarios. One can create 2 targets at the same site for a single source. As you stated if they don't have extra capacity then it becomes challenging.
If you ever go down the xtremIO route, they don't have this same concern as it uses snapshot based replication not journal based.
chrismahon
18 Posts
0
March 1st, 2016 05:00
Thank you, Rich! We ended up switching quite a few to Direct Access as DB batch processing during the DR test was causing the image access log to fill. We did have one CG fill the image access log overnight to 100% and found that it caused the hosts using that LUN to lose access to it. We essentially had to perform test a copy again to the same image which caused some re-work for those hosts. Lesson learned though, don't risk it. Just enable direct access from the beginning.
chrismahon
18 Posts
0
March 1st, 2016 05:00
Ankur, they will be going the XtremIO route next year. I'm looking forward to the DR test next year once it is on XtremIO!
ankur.patel
286 Posts
1
March 1st, 2016 05:00
Sometimes there is a requirement to keep DR replication online while doing these DR tests. I made a video highlighting options. It involves VMware SRM in some instances, but should still give you other ideas if you don't want to go into direct access mode each time.
https://www.youtube.com/watch?v=bHlHz2jk1Fk