I am not sure that I can disable flow control since no switch is present. Is it turned on in the host by default?
-----Original Message----- From: Sašo Kiselkov [mailto:[email protected]] Sent: Saturday, February 16, 2013 6:05 PM To: [email protected] Subject: Re: [OpenIndiana-discuss] ISCSI SSD Performance On 02/16/2013 11:58 PM, Sašo Kiselkov wrote: > On 02/16/2013 11:37 PM, Ian Collins wrote: >> Have you tried the connecting to the volume form another IO host to >> eliminate vmware as the cause? > > Second that. > > The 9k MTU might also be hitting some NIC driver bugs - non-standard > settings can, at times. Since the difference is only in the storage -> > VMware direction, I suspect there may be some trouble there (perhaps > VMware not being entirely okay with how OI is handling the 9k MTUs?). One more thing comes to mind: if all else fails, try to disable flow control. Sometimes it can mess up throughput, if the receiving NIC erratically sends too many TX frames - can be a driver issue in VMware, though this is quite a long stretch. Anyway, right now I suspect VMware might be the culprit here. The number of kittens that have to be slaughtered to get such a high packet switching performance from VMs to storage is astounding. Cheers, -- Saso _______________________________________________ OpenIndiana-discuss mailing list [email protected] http://openindiana.org/mailman/listinfo/openindiana-discuss _______________________________________________ OpenIndiana-discuss mailing list [email protected] http://openindiana.org/mailman/listinfo/openindiana-discuss
