One of the things that I have wondered during preparation for my VCDX defence is why is there a requirement of 622 Mbps for VPLEX enabled metro clusters. According to the blog post by Duncan Epping, the minimum bandwidth required has been changed to 250Mbps per concurrent vMotion for Metro vMotion. Link is http://www.yellow-bricks.com/2012/10/31/bandwidth-requirements-for-long-distance-vmotion/
Here are my questions..
1. to the VPLEX experts .. is the 5ms RTT and 622 Mbps minimum bandwidth requirement strictly for VPLEX Metro and not vMotion requirement from vSphere point of view?
2. One of my colleagues Andrew Best (@AusFestivus) raised a good point while I was bantering over twitter about this, VC doesnt know about the bandwidth between the 2 vMotion interfaces over stretched cluster.. or does it?
3. Following on from that is, If vSwitch has 2 x 10 Gbe uplinks and I have vMotion as well as VMNetwork traffic flowing through the uplinks (logically seperated by VLANs and a QoS applied to vMotion traffic to ensure 622 Mbps availability in the worst case of contention), does vCenter start 8 concurrent vMotions per host even during contention? Or will it go back to the Metro vMotion minimum of 250Mbps per concurrent and only allow 2 concurrent vMotions per host?
I am confused..