I have been digging for a long time now to figure out what the minimum bandwidth requirements are per concurrent vMotion. After a long time I finally managed to get a statement. In the past the statement was made that 622Mbps was the minimum required bandwidth for vMotion, it appears that this is incorrect for vSphere 5.0 and higher. With vSphere 5.0 a new feature called Stun During Page Send (SDPS) was introduced and this has decreased the bandwidth requirements from 622Mpbs down to 250Mbps per concurrent vMotion.
Always nice to know right?!
good to know but maybe the principal limit for the vMotion is the latency value set to 10 ms now.
Duncan Epping says
Not sure what you mean Pablo, the current limit is 10ms with Enterprise Plus license. It is 5ms with the other licenses.
Andrew Fidel says
While that might be technically true I’d caution anyone from thinking it’s a good idea to design a production environment with such little bandwidth for vmotion, even with 1Gbps the SDPS feature brings some of our VM’s to their knees to the point where they might as well have crashed (from the user perspective), in fact it’s why we’re moving to 10Gb next year as right now we have to exclude some VM’s from DRS. To put it in perspective SDPS would be taking a machine with potentially 10’s of GB/s of memory updates and slowing it down to 25MB/s.
This is indeed pure about what it supported. I hardly see SDPS kicking in personally so wonder what is causing it to kick in in your environment. There must be an extremely high page write/change.
This question often comes from people exploring stretched cluster scenarios, where bandwidth typically is a constraint.
Michael M. says
How can I know if SDPS was used during specific vMotion? Via events in vCenter / vmkernel logs?