Slow SQL Recovery - FLR

[thecyborg]

I have a single Veeam proxy/repository:

Dell R720xd
96GB RAM
2 x 6-core Xeon E5-2603v4
RAID1 SSD for OS
RAID6 (12 x 4TB) for Repo
2 x 10GbE LACP to LAN
4 x 8GbFC to Storage

Production storage is a PureStorage Flash Array
VMware Hosts have 4 x 10GbE to the LAN

I routinely hit around 700MB/s during backups. When attempting an application-level restore from a SQL backup to another DB server, everything works but very slowly. The VBK is mounted under C:\VeeamFLR on the destination server and the file copy begins, but I only ever see about 300Mbps (35MB/s) of network traffic on this machine. The iSCSI connection between the destination VM and the backup repo seems severely crippled for some reason. I can run CIFS copies between the VM and the repo at 1GB/s. Is there any way to troubleshoot what's going on?

[Dima P.]

Hi thecyborg,

Can you elaborate what staging SQL Server are you using while performing the item level recovery? Thanks.

[thecyborg]

veeam-repo1: Repository, Proxy
prod-db01: Staging/config SQL server (SQL 2014)
TESTDB01: Test recovery target (SQL 2014)

The staging server is a VM in the same environment as TESTDB01. I don't see any significant traffic between prod-db01 and either veeam-repo1 or TESTDB01. It appears as though TESTDB01 mounts the backup image from veeam-repo1 and initiates a file copy of the MDF.

It would appear that the staging server is only used for browsing the backup, and then the image is mounted directly to the recovery target. Is that correct?