I guess it is time to go out and buy a lottery ticket… ‘cause I was lucky enough to be one of twelve people (aka: non-VMware employees) to get access to the VMworld 2010 Lab environment ahead of time. Holy crap… this is an amazing setup!
For those of you that attended the conference last year, I am sure you are aware of all the problems encountered with the lab environment. To say that VMware has learned from that is an understatement. The scale, redundancy, and content is AMAZING!
Infrastructure (aka – The ‘kit’)
Sadly, the physical infrastructure was not available for us to see. However, Mr. Dan Anderson (infrastructure architect and on the GETO team) was able to provide some insight into how the magic happens.
The environment is divided into management, computer, and storage clusters.
|Management||All of the servers and databases needed to support 12 vCenters, 1 master vCenter, and 12 Lab Managers|
|Compute||4 racks of compute nodes.
– Dual quad core
– RAM ranging from 24GB to 96GB
|Storage||Primarily EMC and NetApp
About 100TB NFS and FC storage
Get this… the overall environment has a RAM footprint of 36TB (yes… 3 6 T B).
As mentioned previously, the level of redundancy is immense. not only are the individual components redundant, VMware has taken the next step and created redundant datacenters. In the event of a failure, the entire lab can be run from the local or remote side. VMworld Labs are connected with dual DS3s (45Mbps each)(burst to 100Mbps). With help from partners (Verizon in Ashburn, VA and Terramark in Miami, FL), two additional/identical environments are setup… with each site being around 90 – 100 ms from VMworld. As of my tour, the environment is running hybrid remote and local. I do not have any idea when/if/why this would change, though.
So, even if the entire Internet became unavailable, the labs could be run from offsite!
Technologies In Use
The usual suspects are in use as far as technologies are concerned:
- vSphere 4.1 / ESXi 4.1
- vCenter 4.1
- Linked vCenter with a master overseeing all vCenter instances
- vSphere Hypervisor (aka – The free one)
- HA, DRS, Heartbeat, Update Manager, Host Profiles
- Lab Manager
- The product only really provides the framework. Instead VMware engineers have been writing new code (via APIs) to enhance the crap out of it. More discussed later.
- View (complete with PCoIP)
The sheer size of the lab environment is such that it requires Moscone Center West facility.
When you get to West, just take a number and wait for your number to be called. A ‘Pit-Stop’ area is setup for your waiting pleasure. Feel free to wait, discuss topics, etc… with your fellow lab rats.
Once your number is called, you will be guided to 1 of 480 stations. Wyse is providing the thin client setup.
The labs are organized into 6 categories (data center, desktop, development, etc…).
One of the first things you will notice is the sweet UI. VMware developers have been working hard to customize the lab experience for us. This definitely shows! Excellent work VMware! The UI is completely custom and written with Adobe Flex. Sharp, smooth, and efficient… everything we could want.
Labs are allocated 1 hour for completion. However, that may be adjusted if necessary (just ask if you need more time. If there is no one waiting for your spot, you may get the time extended).
Last year, to get help, you would need to raise your hand, get a lab guy, ask question, wait for appropriate help if dude did not know the correct answer.
This year, help is baked into the lab. Just click the “Question” button in the UI. This will notify the lab that you need help, your location, and the lab you are running. The person that helps you will know what they are talking about.
Plus, there are 150 experts at the Labs… so, you should be able to have all your questions answered correctly and quickly.
Waiting Time Reduced!
Waiting for labs to be deployed was one of the most frustrating things about the labs last year. VMware did something about that!
Last year, it took 5-7 minutes to deploy a lab. This year, it is no different. However, this is where VMware stepped in.
Prior to VMworld, VMware sent out a survey asking about which labs sounded the best. The intent was to gauge which labs would be the most popular and they could PRE-STAGE the VMs. Sadly, the survey results were inconclusive as all labs were almost equally voted for. So, while the content teams deserve kudos for developing appealing content across the board, the lab teams cannot pre-stage 480 VMs for all labs.
Don’t fret, though. Labs statistics are being gathered and will be used to pre-stage on the fly. Monday morning may involve an extra 5 – 7 minutes.
A quick note about deploying… the lab takes 5-7 minutes to deploy. Deploy is not building VMs, though… that takes seconds. Rather, all VMs and Services need to start as well. This includes 2-11 ESX, Active Directory, and other specific services depending on the lab you are running. Lab guys do not want you waiting for everything to come up while logged in. 5-7 minutes is worth it.
As if the lab content was not enough, they are giving away prices for the most labs attended, completion speed, etc… Prizes include a free pass for VMworld 2011! Sweet! More contests and prices may be announced in the labs room.
Advanced Labs vs Self Paced Labs
VMware has recognized that people like more in-depth lab instruction. To accommodate this need, the concept of an Advanced Lab has been introduced.
These are not Self Paced Labs, so do not expect to get your hands dirty. Instead, lab content creators and subject experts will be walking through the labs, detailing issues, commentary, discussion, etc… This is a unique opportunity to hear from the experts on their specific knowledge areas.
The labs were picked from the Lab Survey sent out… specifically, the top 10 most interesting labs. See, participating in surveys is a good thing.
The advanced lab are on the second floor of Moscone West… in 250 person rooms. Sweet!
Everyone likes numbers and facts. So, here you go:
- Estimated 20,000 hours of lab information presented by the end of VMworld.
- Deploy around 75,000 VMs during VMworld
- Around 100,000 VMs if running with peak attendance
- Compute nodes = 36TB RAM footprint
- Pre-populate labs based on demand and popularity
- 3 data centers (SFO, Ashburg, Miami)
- Smallest Lab: Zimbra
- Largest Lab: Sandbox (aka – Kitchen Sink) (all products; 11 ESX hosts)
- Data center storage: NFS and FC
- Deploy/Destroy around 5,000 VMs per hour
So, the labs look awesome! 30 labs and hours of fun and learning about some sweet products! Cannot ask for much more… well… maybe for multiple copies of your self to ensure you can get to the labs and sessions you would like to see.
Look for me at VMworld! @arctic_99
—Pictures of the Lab Environment—