Very basic Edge question in NSX-T v3

[jeffj2000]

Hi. I trying to learn NSX-T and currently have an NSX-V Cross-vCenter with local egress. I am not a real network person, and my expertise with NSX kind of ends at the logical switch.

So in learning NSX-T I am trying to get more familiar with the Edge, which seems different than in V, well everything seems different. I certainly will not be administrating or in charge of the Edges, but I would just like to know more.

I have read quite a few blogs on Edge, and they all seem to do different or the versions are different, pre v2.4 or not.

So at work I know our Edges have 2x10GB uplinks and each uplink goes to a router with a separate VLAN and there are no standby uplinks in the VDS PG for them. Ie, Uplink1 in VDS would be VLAN100 no standby and Uplink2 in VDS would be VLAN200 no standby.

Now seeing the design doc for NSX-T v2.5 this seems to also be the design.

Now this is where I hope someone can point me to a good blog or something because I am having terrible issues trying to do this. I have named uplinks in the profile and a failover set with no standby. I create the VLAN segments only using the named uplink and when I create an interface on the T0 it just logically seems correct. I can create a neighbor with the 2 interfaces, 2 different VLANs, but the crazy thing is when I pull in the lab 1 of my 2 uplinks, I can still ping both edge interfaces. This makes no sense to me as it seems like I have mapped everything out where the edge interface only goes through 1 uplink. Do the interfaces failover to the other uplink?

Well I don't expect anyone to answer this, I just would like a good place to start knowing how to solve my issue. Thanks,,,

[simonre · Accepted solution]

Hi Jeffery,

NSX-T Edges are very different from NSX-V Edges!  In NSX-T the edges run T1 and T0 distributed routers (DRs) and service routers (SRs).

G Bhat has written a good practical guide to the NSX-T 3.0 edge -- including discussion of latest single N-VDS / multi-TEP design -- I found this very useful:

https://communities.vmware.com/servlet/JiveServlet/download/41703-1-240559/NSX-T%203.0-Edge%20Design...

Cheers

Simon

[Lalegre]

Hey,

I would like to know what you mean by "pull" the uplinks.

Basically on the configuration you are trying to achieve I assume you have two Edges both in the same Edge Cluster and they have their interfaces inside the VLAN Transport Zone and connected to two different VDS Portgroups in Trunk or Access mode.

When you create the uplinks in the T0, they will be created in both Edges but the connectivity will flow depending of the configuration of your T0, if it is Active-Standby or Active-Active.

Could you create a basic diagram so we can understand better? Sometimes is better with a drawing.

[jeffj2000]

Hi. That is a great document! I am sure this will help me. Thanks,,,

[jeffj2000]

Hi. I will do a diagram, but we did get this working by looking at the document that was referenced. By "pull" I mean we were testing shutting down an uplink on the physical hypervisor, we have 2x10GB uplinks we use for the edge. It worked. We are in a T0 active-active. I got confused because it seems the edge interfaces will move to each uplink, which we didn't understand because they are supposed to be pinned to their respective TOR.

We do have it working in the lab, we just need a better understanding.

We are coming from a cross-vCenter NSX-V on OSPF to Federation on BGP. Its all big changes for us. Luckily we have many months to test before we even start planning the upgrade.

Thanks,,,