Re: iSCSI performance guidelines

2008-06-20 Thread Padmanabhan
Do iSCSI request initial TCP socket buffer size or is it left to TCP to determine size ? I did captures using wireshark, it shows always initiator/target opens TCP connection with a window size 56xx/58xx bytes.( small setup, initiator/target PCs with Gig NIC cards connected via Gig ethernet switc

Re: iSCSI performance guidelines

2008-06-09 Thread Mike Christie
Padmanabhan wrote: > Thanks ... any recommendations or pointers on TCP window sizes ? > Still looking and testing and trying to sumarize what I found. I will try to post tomorrow. --~--~-~--~~~---~--~~ You received this message because you are subscribed to the

Re: iSCSI performance guidelines

2008-06-08 Thread Padmanabhan
Thanks ... any recommendations or pointers on TCP window sizes ? --~--~-~--~~~---~--~~ You received this message because you are subscribed to the Google Groups "open-iscsi" group. To post to this group, send email to open-iscsi@googlegroups.com To unsubscribe fr

Re: iSCSI performance guidelines

2008-06-06 Thread Mike Christie
Padmanabhan wrote: > 3. what other parameters have to be considered when implementing > software based initiators and targets ? > I noticed on the initiator side we sometimes get better performance when we set the max command buffer size smaller. The default is really large - something like 51

iSCSI performance guidelines

2008-06-05 Thread Padmanabhan
Hello All, I was looking for general guidelines to enhance software iSCSI performance by optimizing different parameters at these layers of network stack on a host. iSCSI FirstBurstLength MaxBurstLength MaxRecvDataSegmentLength InitialR2T -- No ImmediateData -- Yes TCP layer Receive/send Bu