This morning John Troyer coordinated a bunch of bloggers for a session over at the VMworld 2010 Hands-On Lab facilities in Moscone West. Adam Zimman, Dan Anderson, and Curtis Pope took turns explaining and demoing the lab to us. The lab itself was built as a cloud-oriented system, using software-on-demand and service-on-demand principles, and relying heavily on remotely-hosted equipment in data centers in Miami, FL (Terremark) and Ashburn, VA (Verizon).
The Lab team is really building on what they’ve learned from other years. There are many more labs this year than last, and they’re all self-paced, though there are options for instructor interaction as well if you have questions or want more one-on-one guidance. Self-paced labs means they can do almost unlimited content, and it’s easier to get lots of people through the labs. Last year they had, all totalled, about 7000 lab seat hours. This year they have almost 20,000, with 480 View stations in eight rooms. Dan Anderson, the lab’s lead architect, had some proud things to say about what they’ve done. “The content is killer, the best content I’ve seen yet. If someone sits for four days, eight hours a day, they might be able to get through all of them. But nobody can complain about not having enough stick time,” said Dan.
Perhaps he’s never met some of the curmudgeonly people that attend VMworld. :) But I really appreciate the iterative approach they’ve taken this year to making the labs better. For instance, they learned that pre-registration for the labs didn’t work very well in other years, so it’s all first-come, first-served (FIFO). There’s a check-in station that works with your badge number, and a waiting room with couches and whiteboards and Subject Matter Experts while you wait. The labs will be open from 8 AM until 10 PM every day, too, and they will be offering a prize to “dedicated individuals” (they thought speed and quantity might be the factors, but it isn’t set in stone). They did say the prizes would be something like a pass to VMworld 2011, though, which is very cool.
The hardware and software powering the lab is pretty amazing, with a number of sponsors contributing staff, equipment, and software to make it run. Sometimes on very short notice, too. And in some cases this lab is the largest deployment yet of these technologies. They’re pre-populating lab environments with instances of each lab setup, to avoid the on-demand 5 to 7 minute wait from last year, which is great. They’re worried that they’ll have the prepopulation levels off a little on the first day, but even if you do get caught waiting you can still read the manuals. They estimate that the labs are using roughly 36 TB of RAM (yes, TB) and there’s about 200 TB of storage, between EMC and NetApp, in each data center powering the labs, all connected via NFS. The storage itself is everything from enterprise flash (EFD) to SATA, with the EFD often being used as FastCache to front-end the slower storage.
The stations themselves are Wyse thin clients, with dual monitors and even dual chairs, even though it’s geared for one-on-one learning. It’s all about flexibility and options, which extends to the content itself — the vSphere Sandbox lab is just a deployment of all of their products, for freeform messing around. The Lab team even has redundant wiring to the lab stations, just in case they need it (“We even have redundant chairs!” said Adam). They’re flexible, they’re ready, and they’re hoping that they can set records for the number of happy people in the labs this year. And if you’re not happy, there’s 150 staff floating around to help you out, as well as two lab captains per room.
I’m looking forward to it — labs have always been a highlight of VMworld for me, and these guys are making it even better. I know it’s a lot of work to build, in two months, what usually would be done in a year or two (and then, as Dan said, “throw it on a truck.”). On behalf of all of us, thank you Adam, Dan, and Curtis (and all the others that we didn’t meet). I hope all your hard work is a giant success!