Hello, I'm trying to understand how it is possible to separate vmotion traffic on a distributed switch from other vm traffic generated from VMs. For example, in my test ENV I have 1 host with 4 physical vmnics. When I create a distributed switch and specify the 4 ports it creates the default dvPortgroup with the 4 uplinks. Now, if I create a VMKernel port on a different port group on the distributed switch it's still going through the 4 uplink ports so where is the physical separation? How can I tell one port group(let's say my VMs) to use vmnic1 and vmnic2 and my second portgroup to use vmnic2 and vmnic3? Any help is appreciated. On a standard switch there is that one to one relationship so it's clear how vmotion traffic is using one vmnic and VM port for VM's is using another but it seems like with the distributed switch , it's using all the vmnics avaliable for all traffic, am I correct here? Now I know for one host it would not make sense to use distributed switch but I'm just trying to understand how you achieve that physical separation using a distributed switch?
You can specify Active/Standby/Unused uplinks from within the dvportgroup itself. As always, you'll want to VLANs for logical separation. In addition, you can dedicate physical NICs for certain traffic types in which case you'll define the Active/Unused uplinks at the dvportgroup level. For 10GbE adapters, you'll need some sort of QoS whether that's at the server platform layer (like UCS) or if you have Enterprise Plus, Network IO Control.
All the best,
Mike
Message was edited by: Mike Brown to recorrect stupid autocorrects
You can configure Active, Standby and unsused dvUplinks in the Failover and Teaming policies for each port group.
André
You can specify Active/Standby/Unused uplinks from within the dvportgroup itself. As always, you'll want to VLANs for logical separation. In addition, you can dedicate physical NICs for certain traffic types in which case you'll define the Active/Unused uplinks at the dvportgroup level. For 10GbE adapters, you'll need some sort of QoS whether that's at the server platform layer (like UCS) or if you have Enterprise Plus, Network IO Control.
All the best,
Mike
Message was edited by: Mike Brown to recorrect stupid autocorrects
Thanks Mike, one clarification. you mentioned "You can specify Active/Standby/Unused uplinks from within the dvportgroup itself" and also "dedicate physical NICS for certain traffic types which is still done at the dvportgroup level? it seems like I'm choosing the NICs at the same level(dvportgroup level) for both of your statements. can you kindly clarify from within dvportgroup itself and at the dvportgroup level. Thanks for your help
Sorry - same thing. Im typing from my phone and didn't want to edit.
Thanks Mike.