-
- Service Provider
- Posts: 24
- Liked: never
- Joined: Jan 20, 2017 2:45 am
- Full Name: Mac
- Contact:
Backup Job Questions / Recommendations
We ran a backup job of 10 VMs (1500GB) from a local RAID-5 stand-alone ESXi host to an iSCSI Lenovo NAS over gig ethernet with about 70MBps rate which is great. We decided to setup a dedicated super-micro box for veeam. I know BP guide says to install a physical server for this but the problem here is that it's not able to boot the compute resources during instant vm recovery so we have ESXi running with veeam as the only vm on the host. The drives are 7K 3TB x 4 in a RAID-10. We are seeing performance around 10-20MBps now.
Would we be able to increase performance by installing a proxy on the source host? How much of a difference would this make? The original setup listed bottleneck as source and now it is showing target which doesnt make much sense since this is a RAID-10 w 4 drives - is the issue that we are running target as NTFS on VMDK on VMFS or is that not a performance issue?
Would we be able to increase performance by installing a proxy on the source host? How much of a difference would this make? The original setup listed bottleneck as source and now it is showing target which doesnt make much sense since this is a RAID-10 w 4 drives - is the issue that we are running target as NTFS on VMDK on VMFS or is that not a performance issue?
-
- Veteran
- Posts: 7328
- Liked: 781 times
- Joined: May 21, 2014 11:03 am
- Full Name: Nikita Shestakov
- Location: Prague
- Contact:
Re: Backup Job Questions / Recommendations
Hello and welcome to the community!
Could you provide the full bottleneck statistics?
It`s usually recommended to start working on the weakest part of backup data flow, so if target is an apparent bottleneck, changing backup transport mode will not help much.
Thanks!
Could you provide the full bottleneck statistics?
It`s usually recommended to start working on the weakest part of backup data flow, so if target is an apparent bottleneck, changing backup transport mode will not help much.
Thanks!
-
- Service Provider
- Posts: 24
- Liked: never
- Joined: Jan 20, 2017 2:45 am
- Full Name: Mac
- Contact:
Re: Backup Job Questions / Recommendations
Target is 95%
-
- Veteran
- Posts: 7328
- Liked: 781 times
- Joined: May 21, 2014 11:03 am
- Full Name: Nikita Shestakov
- Location: Prague
- Contact:
Re: Backup Job Questions / Recommendations
What about the Source load for the same job run?
-
- Service Provider
- Posts: 24
- Liked: never
- Joined: Jan 20, 2017 2:45 am
- Full Name: Mac
- Contact:
-
- Veteran
- Posts: 7328
- Liked: 781 times
- Joined: May 21, 2014 11:03 am
- Full Name: Nikita Shestakov
- Location: Prague
- Contact:
Re: Backup Job Questions / Recommendations
General recommendations for target bottleneck is to switch to backup methods with less IO operations such as simple forward incremental backup or forever forward incremental and use faster disks(though looks like you have good ones).
As for proxy server, it`s recommended to have the proxy with direct access to the datastore.
What backup method do you currently use and what is the transport mode for the backup job?
Thanks!
As for proxy server, it`s recommended to have the proxy with direct access to the datastore.
What backup method do you currently use and what is the transport mode for the backup job?
Thanks!
-
- Service Provider
- Posts: 24
- Liked: never
- Joined: Jan 20, 2017 2:45 am
- Full Name: Mac
- Contact:
Re: Backup Job Questions / Recommendations
I changed the proxy so we have one on each host and configured for virtual direct access (we have no shared storage) and it seems to help quite a bit.
-
- Veteran
- Posts: 7328
- Liked: 781 times
- Joined: May 21, 2014 11:03 am
- Full Name: Nikita Shestakov
- Location: Prague
- Contact:
Re: Backup Job Questions / Recommendations
Could check what transport mode is shown in the job "Actions" pane?
Proxy connection type should improve the Source part, while your bottleneck is Target, so I would also change the backup method.
Proxy connection type should improve the Source part, while your bottleneck is Target, so I would also change the backup method.
Who is online
Users browsing this forum: Semrush [Bot] and 39 guests