Identfying Bottleneck Issues

[mcmcomput]

I am having some issues for some new backup jobs to some re-purposed HP servers. Below are my stats.

Load: Source 2% > Proxy 8% > Network 0% > Target 99%

Processing rate 5Mb/s and took 7:36 to replicate 380Gb of which it actually transferred 139.6GB
Source
Dell R710, 2x X5570 @ 2.93/4 cores ea, Perc H700-512Mb w/ 6x 15k 300Gb hard drives in a RAID10. 9VMs w/ 380Gb ESXi v5.1
ProxyDifferent physical server than Source)
Dell R710, 2x X5570 @ 2.93/4 cores ea
Network
900Mb VLS from DC to DR Site (Layer 2)
Target
HP DL180 G6, 2x X5675 @ 3.07/6 cores ea, HP P410-256Mb w/ 8x 7.2rpm SATA 4TB hybrid drives in a RAID10. ESXi v5.5

How can I better identify the bottleneck? Both Veeam Proxy's at both sites are on the same VLan as ESXi. I assume it would be disk, but I can run the below utilities during my backup process and shows I still have plenty of read/write power from the RAID.
AnvilPro Storage Utility
77.09MB/s read for Seq 4Mb / 72.75Mb/s for Seq 4Mb write.

[Vitaliy S.]

What is your backup mode for the backup job (forward or reversed incremental) and what is the Veeam B&R version you're using?

[foggy]

And also what type of backup repository is involved?

I also didn't get "ESXi v5.5" here, is it a host where VM holding the repository resides? Or are we talking about replication jobs?

Target
HP DL180 G6, 2x X5675 @ 3.07/6 cores ea, HP P410-256Mb w/ 8x 7.2rpm SATA 4TB hybrid drives in a RAID10. ESXi v5.5

[mcmcomput]

-Its a replication job from my production servers to a DR server.
-Destination is the DR server / Internal RAID10 storage.
-The Repository for replica meta data is a 100Gb partition also on the same RAID10 destination server.
- Veeam B&R v8.0.0.917 Enterprise

That answer the questions?

[alanbolte]

-The repository for replica metadata should be on the source side, not the destination.
-Double check your job statistics to be certain the correct target-side proxy is in use. What transport mode is listed (hotadd/nbd)?
-If you've been using a physical target-side proxy, try configuring a virtual proxy as an isolation test.

[mcmcomput]

After researching it I think my issue was the HP P400 controller. I had the latest firmware, but i had a spare P812 laying around and after swapping it out get approx 100MB/s on my backup speed with the source as the bottleneck I am happy with these results.

In a month or two i will be sending this equipment to our Dallas data center. We have several TB that is replicated and I have changed the compression to extreme and target to WAN.
Q. Is there any added benefit to have replica seeding turned on after I already have replica's on the target side? Where should the seeding data reside, source or target site?

[foggy]

I still recommend specifying the source side repository for storing replica metadata.

Q. Is there any added benefit to have replica seeding turned on after I already have replica's on the target side?

No.

Where should the seeding data reside, source or target site?

Target.