Strange Question on Performance

[ekisner]

So we usually back up our SQL servers with VAW directly to B&R, which has a LUN presented from a storage array (same rack) over 8Gbit FC. Array was plenty fast, and in every case it was the production array that was the bottleneck (>90%).

As a bandaid fix for our VAW coverage when said storage outright failed, I set up a CIFS repo at my colo, and added it to the B&R library.

Previously, with no throttling involved, the jobs ran reasonably, roughly 170-190MB/s each.

First run to this CIFS repo was on Friday, and the jobs were running between 250-300MB/s each. It was so fast that it saturated my WAN (direct fibre), something I've never been able to do before. I'm not complaining about the saturation, I just turned on some throttling and that is gone. I'm just curious how something that should have been slower was able to go not just faster but WAY faster?

[PTide]

Hi,

I set up a CIFS repo at my colo, and added it to the B&R library.

Which server did you assigned the role of a CIFS-gateway to? Was it VBR or some WIndows machine at the remote colo?

Thanks