EANTC
Op Ed by Mitch Vaughan
“We support that feature and have no issues interoperating with other vendor platforms,” reads the piece of marketing collateral published by a networking manufacturer. But short of going through a lengthy Proof of Concept process, which can be exponentially more difficult to coordinate when involving multi-vendor topologies, how can we validate that statement? It would be nice if there was a vendor neutral organization that hosted an annual event where networking manufacturers could back this statement up.
Enter the European Advanced Networking Test Center: the EANTC.
Every year, this vendor-neutral organization provides a big sandbox for networking manufacturers to connect into and prove two key things:
- The features they claim to support actually work
- They can play nice with others
The protocols and features being tested change from year to year, but the focus is primarily on what would traditionally be considered “Service or Hosting Provider'' technologies: the acronym soup of MPLS, VXLAN, EVPN, SR, SR-TE, VPWS, OISM, etc. However, with technologies such as EVPN/VXLAN making their way into enterprise designs, the testbed is more applicable to an enterprise audience today than ever before.
In late 2021, I was asked to participate as a member of the team representing Arista in the 2022 EANTC event. A chance to participate in what I’ve always viewed as the Network Olympics? Yes, please.
After accepting the invite, the preparation was immediately underway. One of my favorite quotes can be attributed to Louis Pasteur, “In the fields of observation, chance only favors the mind which is prepared,”, often summarized as “Chance favors the prepared mind.” The team adhered to this mindset and went straight to work choosing the platforms that would be used in the event and what software version would be installed on the platforms. We developed the configurations and tested the features prior to shipping the gear across the pond to Berlin where the event is hosted.
At the start of the event, each manufacturer’s team shows up to the Test Center to install and cable up all of their gear. The next two weeks are then spent testing, validating, breaking, testing, adjusting, validating, documenting and testing some more.
In short, the EANTC is an imposter's worst nightmare. A proving ground of sorts that at times can feel like a Rubik’s Cube of interoperability scenarios. A participant doesn’t receive a grade of “Pass” from the EANTC for a scenario simply because the control-plane looks correct; the data plane is also validated. This is done by opening the floodgates on a traffic generator and sending traffic at line rate into the test bed. Once that traffic generator reports zero drops, and the control-plane output looks as expected, then the OEM gets the coveted “Pass” for that scenario. Each participant is validated one at a time by the EANTC engineer/proctor, and the criteria for what counts as a “Pass” is dependent on the role of the participant in the test case. Sometimes a participant may just be a transit or “P” node, while other times a participant may be a VTEP/PE node providing services and connectivity to CE nodes.
When testing so many protocols, features, and scenarios, it’s of course just a matter of time before something doesn’t work as expected. Especially since some of the features being tested are on the bleeding edge.
One might think that when something inevitably doesn’t work as expected, the event would devolve into finger pointing and food fights. But, my experience was that the exact opposite is true. If any participant encountered issues during a test scenario, other participants jumped in offered to help troubleshoot and didn’t throw stones. I believe that part of this was due to the fact that everyone I had interacted with during the event was genuinely cool, but the other part was due to the fact that everyone there knew it was just a matter of time before their stuff broke or didn’t work as expected. Then they’d be the ones asking for help and/or more time.
One of my favorite parts of the event was working with all of the teams involved from different manufacturers. Some may expect that an event full of talented engineers from different network manufacturers would end up turning into a nerdier version of West Side Story in short order, but instead, a sense of comradery existed amongst the group, knowing that we were all there to geek out for a few weeks and prove that we could make all of this cool stuff work together.
We’ve all likely heard the term “you never want to be the smartest person in the room”. Well, this was certainly not an issue for me at this event, and I loved every minute of it. Some of the participants had their names all over IETF drafts and/or RFCs. During one of the tests, I was working through an issue related to EVPN/VXLAN Optimized Inter-Subnet Multicast (OISM) with an engineer from another manufacturer. As we worked through this issue together, I pulled up the draft and noticed his name plastered right there on the author section. The conversation went something like “Hey, it says here in the draft that….wait…you wrote this?” We ended up solving the issue, and we had some fun banter along the way.
After two weeks and hundreds of tests, data captured from all of the testing is summarized and the annual report is created. This report is published to the public and is a great resource for those looking for a means to validate that initial statement of, “We support that feature and have no issues interoperating with other vendor platforms.”
The report for the 2022 event, along with other goodies, can be found here: https://eantc.de/showcases/2022/mpls_sdn_interop.html
Before too long we’ll be preparing for the 2023 EANTC event. Here’s to the network olympics!