Short update for those that read the original blog post: it turns out that the answer to the question “Is it possible to run VMware NSX on redundantly-connected hosts in a pure L3 data center fabric?” is still NO.
VTEPs from different ESXi hosts can be in different subnets, but while a single ESXi host might have multiple VTEPs, the only supported way to use them is to put them in the same subnet. I removed the original blog post.
A huge thank you to everyone who pushed me with their comments and emails to find the correct answer.
I’ll be doing several on-site workshops in the next two months. Here’s a brief summary of where you could meet me in person.
A bit of manual geolocation first: if you’re from Europe, check out the first few entries, if you’re from US, there’s important information for you at the bottom, and if you don’t want to travel Europe or US, there’s an online course starting in September ;)
Does Cisco ACI use VXLAN inside the fabric or is something else used instead of VXLAN?
ACI uses VXLAN but not in a way that would be (AFAIK) interoperable with any non-Cisco product. While they do use some proprietary tagging bits, the real challenge is the control plane.
Does it make sense to run OpenStack on top of VMware infrastructure? How well does NSX work as a Neutron plug-in? Marcos Hernandez answered these questions (and a lot of others) in the Episode 61 of Software Gone Wild (admittedly after a short marketing pitch in the first 10 minutes).
I’m presenting at two Data Center Interest Group Switzerland events organized by Gabi Gerber in Zurich in early June:
- In the morning of June 7th we’ll talk about software-defined security, data center automation and open networking;
- In the afternoon of the same day (so you can easily attend both events) we’ll talk about VMware NSX microsegmentation and real-life implementations.
I hope to see you in Zurich in a bit more than a month!
A few months ago VMware launched NSX version 6.2, and I asked my friend Anthony Burke to tell us more about the new features. Not surprisingly, we quickly started talking about troubleshooting, routing problems, and finished with route-health-injection done with a Python script. The end result: Episode 50 of Software Gone Wild. Enjoy!