Ques on NAS Testbed

[nas.study]

Hi,

I have following questions on test setup used to benchmark various NAS products:

1. Robocopy: This copies 38 files of 1GB to 4KB filesize. If the DUT's DRAM is >= 1GB, would be a cache effect? I have heard that the Robocopy filesize should be atleast 2x of RAM..is that correct?

2. Is it possible to get various files used to run following Robocopy test? I'm curious to run it against our reference platform.

Thanks a lot for your time and help..

 

[thiggins]

1. Robocopy: This copies 38 files of 1GB to 4KB filesize. If the DUT's DRAM is >= 1GB, would be a cache effect? I have heard that the Robocopy filesize should be atleast 2x of RAM..is that correct?

If you want to bust cache, yes, filesize should be larger than the cache on the target platform. But that is not the goal of the benchmark.

2. Is it possible to get various files used to run following Robocopy test? I'm curious to run it against our reference platform.

The files are copyrighted material, so we cannot provide them. It's folder is a ripped copy of Firefly disk 4.

 

[nas.study]

Hi Tim,

If robocopy is copying multiple files in a folder, what are your thoughts on using /mt option (Multi-threading)? In my local env, it seems to help in performance boost.

Please let me know your thoughts.

Thanks.

 

[thiggins]

If robocopy is copying multiple files in a folder, what are your thoughts on using /mt option (Multi-threading)? In my local env, it seems to help in performance boost.

Could help. But that's not how I use Robocopy.

The purpose of the benchmarks I use aren't necessarily to show a product at its peak performance. But to provide a consistent relative comparison among products.