BladeSystem - General
1748176 Members
4234 Online
108758 Solutions
New Discussion

Virtual Connect switches throughput issues

 
gratchie
Occasional Contributor

Virtual Connect switches throughput issues


Hi,

We're having some major nfs read throughput issues with the VC 1/10Gb-F switches. The reads coming off our NFS server to a blade can only go up to 3.6MB per second, where a typical nfs read speed can go up to 90MB/sec.

Has anybody seen this problem too? thanks,


1 REPLY 1
Robert_266
Occasional Contributor

Re: Virtual Connect switches throughput issues

We are having a similar problem. A about 2 months ago our storage vmotions took about 20 min for 10Gb of data. Now they take about 6 hours, although no one is complaining about performance. We have two c7000 enclosures with 27 BL460G1 servers with the same problem. We are using NFS storage on 2 Netapp FAS3170 clusters with 10Gb links. Links are not saturated anywhere. Vmotion works fine on new enclosure with Flex-10 vc and vsphere, so not storage. We just built empty enclosure with same vc swithes and updated all firmware. Farm is ESX 3.5 update 3. New server in empty enclosure is ESX 3.5 update 5. Will post if we find something.

RJ