We are considering adding DAS to existing systems. However while benchmarking current set-up I noticed the sequential write/read throughput was considerably lower than what I expected.
SCv2080 with 82 8TB NLSAS. Volume profile: balanced (RAID10 DM+RAID6). Front end is connected to R730 running CentOS 7.5 with 12Gbps SAS.
To make sure everything is set up correct, I created a new volume and format with XFS per instructions in SC-Series-with-RHEL-7x-Dell-EMC-2018-(CML1071).pdf
I then used iozone commands provided in Dell HPC NFS Storage Solution WP.pdf and I was getting merely 600MBytes/s, with 82 spindles. That sounds nowhere near normal to me. Do you guys have any insight into this and what is the advertised throughput for SCV2080 btw?
With your SCv2080 they are traditionally setup to deal with the high IO. When it comes to throughput you will max out after 10 drives. I will send you a private message & if I can get you to send me your controller serial# for your SCv2080 we can look at your configuration to ensure that there are no issues & all is setup correctly.
Please let us know if you have any other questions.
On a side note, how about ME4084? I wonder if it may be more optimised toward throughput performance? We may be purchasing a few of them but judging from the appearance they come from the same (hardware) OEM partner as SCv2080.
DELL-Sam L
Community Manager
•
8058 Posts
•
34386 Points
847
0
Posted July 16th, 2019 08:00
Hello 5X5CX62,
With your SCv2080 they are traditionally setup to deal with the high IO. When it comes to throughput you will max out after 10 drives. I will send you a private message & if I can get you to send me your controller serial# for your SCv2080 we can look at your configuration to ensure that there are no issues & all is setup correctly.
Please let us know if you have any other questions.