Notes from an Official Record Breaker!
Infinera Product Evangelist
When I was a kid growing up in 1970s Britain, one of my favorite shows was called Record Breakers. It was all about Guinness World Record attempts, and a lot of the records were broken right there in the TV studio.
Since I was always the last one chosen for any team sport, I always assumed that my only chance of becoming a record breaker myself was to hang onto the coattails of an amazing technology that I could deploy in some kind of record breaking scenario.
A few months ago my opportunity finally came along. We’d just completed the 2 Tb/s in less than 12 minutes demonstration on the GÉANT network with DANTE, and the idea was raised that we could maybe get this “rubber-stamped” by the Guinness World Records (GWR) organization.
Unfortunately GWR doesn’t just rubber stamp record attempts…they actually take their role pretty seriously. First we had to persuade them that this was a valid record, and that it would be possible for other companies to try to break it. I put together a test specification that was amended and approved by GWR, and I spoke to several of our customers about the idea of using a live network link to host the record attempt. Once again, DANTE stepped up to the mark and offered a number of possible routes for us to use. One of the snags was that the GÉANT network rollout was almost complete, and there might be understandable resistance if we started rerouting traffic right after the links had been set up.
But finally a route was decided: from Amsterdam to Hamburg. We’d already used the Amsterdam PoP for the 2 Tb/s trial, and it was good to be working in familiar surroundings, and also with a really great bunch of folks who run that PoP.
The next challenge came when we decided how much capacity we would turn up. We’d already shown 2 Tb/s…so how about 4 Tb/s? But why do 4 Tb/s when we could potentially do 8 Tb/s? Once we’d stopped chuckling about the total insanity of a trial on that scale we looked at the question again. Why not do 8 Tb/s?
If you’re an Infinera follower you might remember we ran a trial of a 10 Tb/s super-channel along with Telefonica in their test lab in February this year. That was an amazing test, but it was done in a lab, not on a real network link, and with prototype equipment. The 8 Tb/s test we were now thinking of would be on a real network link, using production equipment, and racing against the stopwatch to turn up the capacity. All in all, quite a challenge.
I half hoped that when I asked Sivaram Balakrishnan, the man in charge of our demonstration projects, if he could muster a full 8 Tb/s of equipment at both ends of the link he’d say I was crazy…but he didn’t. Instead over the next few weeks he worked on getting his hands on all of the gear I needed.
All through the process the DANTE CTO, Michael Enrico, was extremely supportive, as was the entire DANTE team. All the pieces were falling into place, and now we just needed to figure out if this could be done against the clock – in other words, a provisioning speed test.
The week before the record attempt we had all of the equipment in place in both Amsterdam and Hamburg; we had our independent adjudicator, Wolfram Pueschner, and our Guinness World Record official adjudicator, Sam Mason, all booked for The Big Day.
On the evening of Thursday, August 8, five days before the record attempt, we were given a five hour planned maintenance window in which we could get all of the equipment transferred from its delivery boxes in the PoP and into the racks. We had to reroute the existing traffic away from the Amsterdam-Hamburg link, and we then had to actually do the speed provisioning test itself. It was just before midnight that we all found out for the first time that this could actually work. All we had to do now was replicate it in front of the two adjudicators the following Tuesday.
At this stage I also want to mention our installation partners, UCS. Andre Diac was the engineer on site who helped us complete the installation and stay in the game. I asked him if installing the Infinera equipment was easy. “Too easy!” he laughed. “Sometimes it makes me feel like a technician, not a network engineer! I work with many different vendors’ equipment and Infinera is totally different. Totally easy to use.”
On the day of the test, while my colleagues Steve Pegg and Kevin Elliot were doing the hard bit and getting the gear ready, I briefed Sam Mason from GWR while Wolfram Pueschner looked on and made sure our procedures were solid.
Sam asked Wolf when he thought another vendor might break the record if we were successful, and Wolf just shrugged. “Who knows? The key to the speed of Infinera provisioning is the super-channel. Without super-channels nobody can match or beat this record – it’s as simple as that. Some vendors are talking about 400G super-channels coming soon, but I didn’t see anyone shipping anything yet except Infinera.”
Steve and Kevin said we were ready, but then Wolf told us he had a suggestion.
“You know…if you want to simulate a real operational deployment you should have only one engineer plugging the line cards.”
Steve, Kevin and I looked at each other, trying not to panic. Even though the Infinera line cards have five times the capacity of a 100G transponder, we still had 16 line cards to plug in at the Amsterdam PoP. We’d actually planned for four of us to plug in the line cards in a kind of “operational ballet.”
I turned to Sam and asked him for advice, hoping he’d say no – after all, he was the guy representing GWR. He thought about it for a second and gave his judgment. “You’ve got a detailed test plan that we approved. As long as you comply with the test plan we will award the record. But if you’re telling me you want to go beyond the test plan and increase the difficulty then we will still award the record. We can discuss it, and amend the test plan for future record attempts.”
OK…so our well-planned team effort of plugging in the line cards was scuppered. We now had only one engineer allowed to touch anything. Suddenly things had gotten a lot harder.
We decided Steve should be that man…and we set to work. To give you an idea of how this works, at each end of the link there are two DTN-X chassis units, and these are clustered so they appear as a single node. Each chassis has 10 slots, and we had a total of 16 line modules that had to be plugged in, and the fiber connected into the line system. All this by just one engineer.
The clock started when the first line card was inserted. Steve plugged in the remaining 15 line modules as quickly as he could, and we specifically noted which was the last line module to be inserted. Then he started fibering up the boards.
For the next ten minutes or so we waited while the line modules booted and ran through a series of automated configuration protocols. Then the first super-channels started to come up on the Optical Spectrum Analyzer (OSA) screen. Maybe I need to get out more, but I thought it was exciting!
Despite the long amplifier chain between Amsterdam and Hamburg it’s just incredible to see how quickly the super-channels were established.
As soon as the final super-channel was up, we needed to demonstrate that it was “digitally available” for services to use. This is why we needed to note which line card was the last to be inserted, because we knew this would be the last one to boot and come up. With the clock still running, this was the super-channel over which we chose to provision a 100GbE service. Kevin used the “drag and drop” provisioning capability in Infinera’s management application to make that happen. I have to say I couldn’t help teasing him that somebody had managed to create a 100GbE service in 9.15 seconds at our recent London Office opening competition. Sadly I can’t repeat what he told me to do at that point.
We knew the 100GbE service would be routed over the final line module because Kevin had already set up the Traffic Engineering weights in GMPLS so that it would be forced to use that path.
And bingo! The 100GbE Test set screen turned green, and Sam stopped the clock. 19 minutes, 10 seconds on our first attempt!
All of us clustered around the stopwatch. “We can do better than that!” Steve said.
We took a quick break while Kevin shut down all the line modules and pulled them and the fibers out for the next run. 19 minutes, 55 seconds! Slightly worse. Steve gave some weak excuse about the catches being stiff on one of the boards, and we went again. Our last chance…19 minutes and 1 second! A full 9 seconds shaved off our first effort.
“Gentlemen, you have a new world record!” Sam said, grinning.
It felt a bit odd. I’ve been living and breathing this project for almost four months. A whole team of amazing people helped me make it possible and finally, after all these years…I’m a Record Breaker!
Thanks to all who participated in setting this record.
Please see Press Release for additional disclaimer information.