Veeam slow suddenly

[sven.hanke@kwenet.de]

We have installed Veeam on one physical server (W2K8R2, '16GB Ram Raid 5 with Veeam BackupRepository = 5 * 2 TB SATA)
and have made ​​a backup of a VM.
First a full backup at 25 MB / s without any adjustments
Second an incremental backup with 255 MB / s with no adjustments

now we have only 5MB / s for inkremental

the 8 Gbit Fibre Channel adapter has minimal load

I've created a backup proxy manually assign the DataStore

Veeam reports only:

:: Load: Source 99%> 51% proxy> Network 0%> Target 0%
:: Primary bottleneck: Source

The two ESXi's minimum load we only have 7 VMs.


Does anyone have any idea?

[foggy]

Sven, what transport mode is effectively being used by the proxy (and which was used previously, during the first and second job runs)? You can see it when selecting the particular VM to the left in the Realtime Statistics window (to move between job sessions, use arrows).

Do you see that behavior for all your VMs or for the specific VMs only?

[dellock6]

In addition, to be sure SAN mode is beeing used, you can force this by editing the proxy settings, so at least transport mode can be excluded from the possible reasons of slowness (I think Alexander is suspecting an unwanted but possible failover to network mode...)

[sven.hanke@kwenet.de]

Hello dellock6,

in the configuration i have checkt the direct San Access and uncheckt the failover to network mode. the backupproxy is the server who i have installed the Veeam he has an 8GBit FC HBA to the SAN.

[sven.hanke@kwenet.de]

today i take a look on my VeeamServer, the SQL-Server of Veeam has 70% off CPU load from my 7 Jobs are 5 standing on detecting 2 Jobs are running very slow with 374 kbit.
after i have stop the jobs the CPU load is normaly

[dellock6]

Are you using internal SQL Express service (that installed during Veeam setup)? I've got too some performance problem with SQL slowing down backups and eventually crashing some of them, but in my scenario Veeam is in a VM with 6 Gb ram and I'm saving hundreds of VMs, so nothing comparable with you.
What are the bottleneck stats now with explicit san access settings?

[sven.hanke@kwenet.de]

the bottleneck is with 95% Souce, i think as souce it means the proxy and as proxy is configured my server how i have installed Veeam, BackupExec, Veeam One but in other installations this works fine.

[dellock6]

No, source is the source storage you are extracting vmdks from, proxy is listed apart as you can see
Uhm, pure guess, are there any cuncurrent job running on the datastores that are maybe slowing down the storage? FC fabric is correctly configured?