high read latencies when backing up with 3PAR integration

[yunales]

Hello Veeam community,

I'm experiencing the following issue with our Veeam + production environment:

we are experiencing high read latencies(100-300ms) when SAN backup is active with our HPE 3PAR integration. The latencies can be seen on ALL production VMs and on ALL esxi Servers(50 +) when using 3par commands vstatlun the latency can be seen also.(frontend latency)
vstatvv on the other hand is at 4-5ms.(backend latency)
We can see it since the upgrade from 9.0 to 9.5u1.(two weeks ago) Our HPE 3PAR was upgraded to the latest release 3 weeks ago and is currently running on version 3.2.2 MU4. all backups are disabled currently because it's breaking production vm availability! is there a known issue or hotfix for that?

SR number #02131145

is this a known issue? anyone else got the same problems?
A SR was opened with HPE since the blades,san switches are all from hpe.

regards

[jmmarton]

Have you enabled Backup I/O Control (storage latency control)? That could help, especially in light of our advanced data fetcher introduced in v9.5 which accelerates reading data when using BfSS (your 3PAR integration).

Storage Latency Control
https://helpcenter.veeam.com/docs/backu ... tml?ver=95

Advanced Data Fetcher
https://www.veeam.com/blog/enterprise-s ... suite.html

Joe

[Gostev]

Just reduce concurrent tasks on your proxy or repository by half for a start, as it looks like you are simply throwing way too much load at your SAN with current amount of parallel tasks. As Joe already noted, 9.5 backup proxies are way more efficient in data retrieval than in previous versions.

[yunales]

hello all,

thanks for all the replies. Just wanted you know that in the end the interswitchlink seemed to be saturated and all kind of latencies occured on all host due to that.

We fixed that and added an additional backup proxy to optimize the environment. We are using the Storage Latency control feature too now, which seems to help a little.

regards