Hands-On Labs at TechEd 2011 | How does it all work?

This week at TechEd 2011, an attendee favorite are the Hands-On Labs. I met with the good guys from XB Velocity to get a behind the scenes look at how everything works.

At the highest level, the labs are stored in an on-site private cloud and then accessed through Internet Explorer on each PC in the lab. Lets dive in and look at things at a deeper level.

What’s the hardware involved?

What’s in each blade?

  • 2 physical processors each with 6 cores (Intel Xeon X5670, 2.93 GHz, with Hyper-Threading)
  • 128GB of RAM
  • 2 146GB hard drives
  • 1 320 GB HP IO Accelerator Card, powered by Fusion IO (provides 150,000 IOPS, Read speed 750MB/s | Write speed 550MB/s)
  • Windows Server 2008 R2 Datacenter SP1 w/ HyperV

How’s the infrastructure managed?

  • System Center Configuration Manager 2007 R2 – Used for reporting and deploying updates to lab clients
  • System Center Operations Manager 2007 R2 – Used to monitor all servers involved
  • System Center Service Manager 2010 – Used to run help desk and provide incident routing during the show
  • System Center Virtual Machine Manager 2012 Beta – Used for monitoring lab VMs and managing  / monitoring infrastructure VMs

The Nitty Gritty:

The hands on labs this year are all running on a private cloud setup using the aforementioned hardware and software. This private cloud serves 350 users located in the lab area. All of the base lab VHDs (Virtual Hard Disks) are stored on the SAN, and differencing disks that point to these base VHDs are stored on each blade’s IO Accelerator card. When someone wants to start a lab, a copy of the virtual machine (already in a saved state for quick starts) is made, pointed to a differencing disk on the IO Accelerator, and the VM is then turned on and available to the end user.  By doing it the this way, the labs are able to be run at optimal speed and there is little risk of any damage to the Base VHDs stored on the SAN. For better memory management within the VMs, Dynamic Memory is enabled to allow the VMs to take more RAM if they need it, and to give up RAM they don’t need.

End-User Experience:

When the user sits down in front of a lab station they are presented with a list of categories and labs to choose from. When they find the lab they click the Take Lab Now button and that begins the backend VM creation and startup process. Once the VMs have launched, another window opens on the second screen with the lab manual. If there are multiple VMs involved, there is tabbed navigation between each of the lab’s VMs. After the user finishes the labs, the VMs are shut down and then destroyed, or if a user leaves without ending their session, the system will automatically destroy the VMs after 40 minutes of inactivity. I’ve included some screenshots and photos of the setup below.

IMG_0021 IMG_0014

The Serverquarium at TechEd 2011                          Lab menu on left, lab manual appears on right

IMG_0016 image

Running lab at TechEd 2011                        Screenshot of the Lab selection menu

 

image IMG_0032

Screenshot of running lab showing multiple VMs                          Hands-on Lab Private Cloud

IMG_0030 IMG_0028

Inside of one of the blades                                       320GB HP IO Accelerator Card

Blade Enclosure SCVMM2012(beta)

SCOM diagram of BladeCenter enclosure              Hyper-V performance metrics

HP IO card perfmon mosaic

HP IO Accelerator Card Performance                  Mosaic showing all running VMs in HOL area