The datacenter is rapidly changing as we shift to vitalizing many of our application and Web servers to virtual workloads. This is a forum to discuss physical design changes from the “one box, one OS” to various options like blade heads (build out), 16-way heads (build up), and SAN storage of disks. Topic starters like “NFS, LUN, or CIFS for virtual hard drive storage?”, “Why NOT to use iSCSI?”, “What do you require before considering the jump from VMWare to Microsoft for hosting or management”, “Anyone using Citrix over the others?”, “Anyone aggressive with SCVMM and/or Hyper-V yet?”. Hear big and small shops describe what works and what doesn’t for them; how they address F.U.D., bottlenecks, disaster recovery, backups, and more. This discussion is product agnostic and focused around how we can do more OS’s with less hardware. We had this BOF last year—come by to talk about how much has changed.

• Nearly all large production environments were using VMware ESX and VMotion
• Most of those running VMware are testing Virtual Server 2005 R2 to see if they can replace VMware (cost is the big motivator); but the feature set for uptime, redundancy, and performance (SMP) is not ‘there yet’ with Microsoft’s current offering.
• Only a few are testing betas of SCVMM, and the crowd was confused on what features are in SCVMM vs. RS SP1 vs. Windows Server Virtualization.
• No one was building host servers larger than four sockets (they are building out, not up)
• A few are having SAN disk performance issues, even on Fibre Channel. Too many servers patching, booting, etc. can quickly affect all guests on SAN storage
• Costs are not cut and dry. It seems SANs are now required (shared storage) when they weren’t in environments using mostly local storage on servers, which is just shifting their cost from local host (cheap disk) to SAN (not so cheap)
• Most either have iSCSI or are planning for it. One mentioned they are using it for test/dev and only step up to Fibre Channel in production to keep costs down. (I wanted to ask about anyone using iSCSI on 10GbE, but didn’t fit it in)
• Backups are almost always being done via each guest (one agent per guest) but I learned that NetBackup may have a cheaper licensing model for virtual machine guests
• Many were confused or unaware of the Microsoft OS licensing changes around virtualization
• Didn’t seem that many knew of Microsoft’s approach to the .vhd file standard as a format that will eventually be universal (Vista image backup, WUDSS LUN’s, etc)

Thanks to INETA and Culminis for helping making BOF’s happen. Thanks to Allen Stewart from the WinCAT Team, SCVMM Team, and others for answering my virtualization questions on the TechEd TLC floor. I’m hoping we can have the same or similar BOF session next year, and see how Windows Server 2008 Server Virtualization and SCVMM being released will change the landscape.

Trying to keep in touch with your coworkers onsite is still amazingly tough in 2007… Windows Mobile Phone batteries don’t even last through lunch with Communicator & Outlook running; and WiFi is still hit or miss (assuming like past years that it doesn’t reach the dining hall nor the front of the Breakout Sessions). Funny thing thou: if you’re using Communicator Mobile 2005 on your Windows Mobile phones, you can tell when breakout sessions get out, as everyone jumps back online after attempting to save battery during the session… which is a great opportunity to use the person-specific “Tagging” in Communicator, where you can mark someone (unbeknownst to them) so you’ll get the usual “online/offline” toast (pop-ups) but ONLY for the people you tag.

Here’s another TechEd Communicator tip: make a new group in the full client called “TechEd”, add your onsite coworkers, and it will sync up to Communicator Mobile (may have to sign off/on mobile). Now when in Mobile, highlight the group and blast an IM to the group name in just a few steps. May I recommend: “Lets meet at the Birds of a Feather session number 65 tonight at 7:45pm on Virtual Datacenter’s”… or in IM txt: “cu@BoF65@7:45”

Topics of discussion at the 7:45pm event include:

• Why do you use virtual machines
• Biggest issues you’re facing today in using more virtual machines
• Individual virtual server hardware design
• VMWare vs. Microsoft, what made you choose
• Backups and disaster recovery designs using virtual servers
• Identifying virtual machines on the network
• Controlling virtual machine growth and resource overutilization
• Preventing unnecessary “temp” guests from being “always on”

I hope to see you there, and learn from your experiences.

I’ll be moderating this discussion which I hope will help dispel some myths, expose some potential hang-ups (see Breakout Session SVR240 Avoiding Virtualization Gotchas the same day at 1pm in S230 E), and educate us all on how we can better plan and prepare (and hopefully jump onboard) for this next wave in server build-out and datacenter design.

Several ways I this are:

• Subscribe to the RSS feed of TechEd 2007 Blogger’s at techedbloggers.net
• Watch the teched and teched2007 tags at Technorati

This year is different for me in that I won’t be there until Tuesday morning, so until then I’ll be glued to Virtual TechEd.  Not sure if the presenters will be uploading their content to the Schedule Builder on the fly, but it would be cool if I can catch some PowerPoint’s before I get there.

Soon SCVMM will hit the streets, Longhorn will ship, and if you’re a smart Windows shop and implementing SCCM, SCOM, SQL Reporting Server, and the other System Center products, you’re datacenter will be humming along with near-amazing agility and workload insight previously only seen in very highly maintained 1%’er shops. The amazing thing is if you’re willing to focus on System Center products rather than a mix and match of everything from other venders? you’ll end up with something much like a true distributed computing environment with deep insight into what’s going on at a application and “app system” level rather than the more common ‘pinging port 80/443′ to determine if services are up.

I believe this shift to virtual web and application servers will happen so quickly (for those that let it) that I’m hosting a Birds-of-a-Feather session on in at Microsoft Tech-Ed 2007 this year (Tuesday night). Come see what others are already doing to ensure their datacenter is “virtually ready”.

* More than 1,250,000 pieces of Mikes and Ikes were consumed over the course of the week.
* 18,750 pounds of salad were prepared and offered at meals.
* 83,700 ice cream novelty bars and fruit and yogurt bars were served.
* 60,000 (or 5,000 dozen) eggs were eaten by attendees at breakfast.
* It took four tractor trailer trucks to transport the 150,000 bottles of water that were consumed this week.
* The total amount of fruit ordered for this week would fill three-quarters of a full-size tractor trailer.
* At least 1.6 million ounces of coffee were poured.
* More than 50,000 pounds of carbohydrates were consumed at Tech·Ed (Atkins who?).
* 7,500 table cloths were used and reset on a daily basis.