[Bioclusters] Re: file server for cluster

Goran Ceric bioclusters@bioinformatics.org
Tue, 23 Apr 2002 12:59:00 -0700


The numbers make sense. I have a couple of Dell 210S (12 x 73 GB) arrays
attached to 755N's with dual 1 GHz CPUs and 1 GB RAM. Bonnie++ gives me
~30-35 MB/s read/write (2.4.17 kernel, 2 GB file size) with write speeds
being somewhat lower because of my RAID 5 (AMI megaraid) setup.

Goran Ceric
System Administrator
Washington University, St. Louis
Department of Genetics, Eddy Lab
goran@genetics.wustl.edu
http://www.genetics.wustl.edu/eddy


-----Original Message-----
From: bioclusters-admin@bioinformatics.org
[mailto:bioclusters-admin@bioinformatics.org]On Behalf Of Ivo Grosse
Sent: Tuesday, April 23, 2002 9:19 AM
To: bioclusters@bioinformatics.org
Subject: [Bioclusters] Re: file server for cluster


I just read that the throughput of the MaxAttach 6000 (with an IDE
RAID) is about 180 Mb/s, and that the throughput of Dell's PowerVault
755N (with a SCSI RAID) is about 230 Mb/s.

http://www.eweek.com/article/0%2c3658%2cs%253D702%2526a%253D21465%2c00.a
sp

Also, I heard that Dell's new PowerVault 755N based on Tualatins can
reach 470 Mb/s.

Naively, I had thought so far that striping over a dozen of disks gives
throughputs greater than 1 Gb/s.  Could you please comment on whether

- the above examples are poor performers, and typical NAS servers give
a throughput greater than 1 Gb/s?,

- it would make any sense to connect more than one Gigabit line from
the cluster switch to the file server?,

- which NAS servers you would recommend that give a throughput of 1
Gb/s?, and

- how much those beasts would cost?

Thanks!

Ivo

_______________________________________________
Bioclusters maillist  -  Bioclusters@bioinformatics.org
http://bioinformatics.org/mailman/listinfo/bioclusters