Hey all have a few questions for setting up NSX-T in a lab.
I have only 3 hosts to play with in lab. vDS is in use and have 2 10GbE interfaces per host cabled to a single Nexus 9k TOR switch.
I've been reading the early access validated design document: Early Access: VMware Validated Design for NSX-T in a Workload Domain
but still have a few questions. I am used to NSX-v and everyone says forget about NSX-v when thinking about NSX-T. Still struggling.
1) I have a "management" VLAN (2) where vCenter lives. I'd like to put NSX-T Manager there also. When it comes do deploying controllers I plan on deploying 3 controllers in my "management" VLAN(2). Does management VLAN need any L2/L3 connectivity to TEP?
2) Validated design shows "Number of Physical NIC Ports" for TEP as "2". Can these be the same two that I am doing all of my other traffic on? Obvious points of contention without NIOC but can I still utilize ESXi hosts for compute (and N-VDS in this case) that only have 2 uplinks total?
3) As above, do I HAVE to migrate vmkernels to N-VDS?
4) Does my overlay transport VLAN (in this case a single TOR switch so, L2 really) need any L3 access to "management" VLAN(2) where vCenter, NSX-T Manager, and controllers reside? I believe adjacency is needed only for Transport Nodes (ESXi here) to communicate Geneve encapsulated data but I want to be sure. (oops realized this is the same question as #1, but I'll leave it anyhow)
Thanks all! I may add more as I comb through the docs further!
