Unsolved
This post is more than 5 years old
1 Rookie
•
23 Posts
0
1366
February 25th, 2015 05:00
RecoverPower For VM - Journal Question
Hi Guys,
Just need clarification around Journals when using recoverpoint for VM....
Please let me know if the bullet points I am going to list here are accurate or not.
- Each consistency group requires a production journal and a copy journal
- It is recommended that the production and copy journal be the same size
- With recoverpoint for VM, one "Datastore" can hold the "Journals" for multiple consistency groups. That is why it is refered to as the "Datastore pool" when it is initially registered.
- Journal sizing formula:
Journal size= (data per second Mbps) * (required rollback time in seconds)/(1- target side log size) x 1.05
- when doing an initial sync (i.e during initial first time replication sync) Journals do not come in play and therefore journal design does not have any impact on the performance of the initial sync.
- Journals "DO NOT RUN OUT OF SPACE" as snapshots are overwritten.
Thank you.
No Events found!
Idan
675 Posts
0
February 25th, 2015 05:00
Hello there,
The production and target journal can be of different sizes. Journals can be as small as 10GB. The requirement on production journal will depends on the retention the solution need to have after failover (when the production journal becomes the target one).
the datastore housing some or all of the journal VMDKs can be dedicated or shared. The decision on whether to dedicate it in general or b/w CGs will depend on the production workload and consequently the target journal required performance.
Correct on the fact that during a full sweep after the CG is enabled, the journal is bypassed, writes are going directly to the target volumes. This is called fast first-time initialization. Although target journal performance isn't related to the performance of an initial sync, it is vital to perform a proper sizing exercise to make sure that required RPO can be met.
Journal are being used in a cyclic manner so old PiTs will be overwritten by new ones.
Hope that helps,
Idan