Things have been pretty quiet around here lately, haven’t they? So it would seem.
I’ve been playing with new toys, old toys, ditching old code, squirting out new code, and learning new things.
Virtualization
My day job affords me some great opportunities to play with the latest and greatest technologies. One of those things is virtualization.
Our shop is mainly VMWare based. We are currently around 65% virtualized, with 51 total hosts on three sites, containing around 1500 virtual machines. Most of these VMs are servers. However, there are approximately 200 workstations in there as well, for development and testing.
VMWare has a pretty solid product, which has been used for years and years by companies around the world, with great success. Recently, VMWare has introducing pricing changes which will greatly affect the budgets of IT departments everywhere. This (along with the declining level of support) has caused a lot of us to re-evaluate our virtualization strategies.
I’ve recently embarked on an adventure with Microsofts HyperV. Hold on, hear me out.
After the big VMWare backlash earlier this year, I began reviewing products and strategies for leaving VMWare. I checked out Xen, HyperV, and even some open-source solutions. The first thing I looked at was support, mainly what we get for our buck.
While the open source community has some really great packages out there, getting support for open source software is typically forum-based. When I have a problem, I need to get someone on the phone now, not post to a forum and wait a day or two for a reply.
I won’t bore you with the rest of the details, but suffice to say, we landed on Microsofts HyperV. We already have a premier support agreement, as well as enterprise licensing. I rarely have a problem with the support process, and if I do, my TAM is right on the ball.
Currently, I’ve built 1 four node HyperV cluster (infrastructure testing), 1 stand-alone HyperV server (pilot), and am in the process of building 1 more four-node cluster (lab testing). All these items are managed by System Center Virtual Machine Manager 2008 R2 SP1.
There has been a bit of a learning curve, but nothing too troubling. This has been mainly related to translation of things I would do in VMWare into Microsoft language.
For a brief period, I even had SCVMM 2012 installed, but decided that it was still a bit too dodgy (it’s currently in beta) to manage any more than a few hosts.
Things have been going swimmingly with the HyperV installation. There are currently about 30 VMs running on the infrastructure cluster, with a mix of database servers, “regular” servers, and workstations. All co-exist quite nicely on the cluster. One user even commented that performance was better on the VM than his local workstation.
Overall, I’m pretty dang happy with HyperV. If you haven’t tried since the early days, give it another look. HyperV has matured quite nicely.
Monitoring!
In large enterprise environments, it is absolutely essential to have a monitoring solution in place. There are many out there, some work great, some, well, not so much.
In our environment, we utilize a few different packages, depending on what we are monitoring. For Windows machines, System Center Operations Manager 2007 R2 SP1 (SCOM) is our tool, while we use vFoglight for the virtual infrastructure. I won’t touch on the vFoglight piece here, though.
SCOM 2007 can be a really powerful package. There are literally tons of items that can be monitored and alerted for. Health status for not just individual machines or services, but for distributed applications. Dashboards. Way too much stuff to go into here.
As of late, I’ve been limping along the current SCOM implementation that I administer. It wasn’t built in a manner that could be described as “measured and thoughtful”. But it works. Unfortunately, it has to be baby-sat.
So, SCOM 2012 is coming out soon, and I’ve been building a lab implementation of the beta. Oh man, are you in for a treat. It may be that fact this install is planned and architected, rather than thrown together, but oooooh baby. I never thought I would get turned on by a monitoring solution. The web console alone is worth writing home about. Not to mention the improved dashboarding. My boss is loving the dashboards!
Another really great thing: No more RMS. Let me say that again: No more RMS. In the past, in a multi node SCOM implementation you had one server that acted as the Root Management Server. When the RMS is down, you can’t perform any admin tasks, no consoles will open (they connect to the RMS), connectors will not function, and notifications will not happen since subscriptions depend on the RMS.
Instead, all management servers will run the SDK service, and all management servers will be able to fill the role of tge RMS. Nifty!
Site Related Stuff
I recently began an improvement effort for fortypoundhead.com. As you can tell, the site got a facelift about two weeks ago, in an effort to have more information presented on screen at once. One thing I don’t like: having to scroll around for information, and this Windows 8 inspired design seems to help with that.
Speaking of Windows 8, I have gotten my copy of the Windows 8 Server developer preview, so watch for tips, tricks and info around the upcoming OS.
I’ve also added public profiles to the site. The data was already there, it was just a matter of securely exposing the data, and adding links here and there throughout the site. A complete user list can be viewed by clicking the Userlist link at the bottom of the page, or if you are viewing a section which includes user interactions, you can click on the users name to see their profile. Cool, eh?
Coming soon are:
- Private Messaging
- Questions and Answers
- Improved mail handling
Private messaging is just that; private messaging between users on-site. Question and Answers will be a place to ask specific questions about a problem you are having, rather than having to tag on to a post thread. Other users will be able to post answers to your questions.
The email handling improvement is mainly a back-end improvement. The site currently utilizes in-line scripting to send emails, mainly for new users registrations, admin notifications, and other small things. With the added load of messaging and the questions module, as well as some other mail-enhanced features that I am planning, it will be necessary to build an external mail handler to take care of the task. Basically, rather than inline mail handling, the site will drop notifications and things to the database, and the external app will poll at intervals for notifications. When it finds notifications that need to be sent, it will lift them from the database and send them off via SMTP. Cool, eh?
The site now contains 14,943 lines of code, within 263 files, and will be growing. No wizards, plugins, or other weird crap. All hand typed, and are my fingers aching.
BUT if there is something you want to see here, drop me a note through the contact form, and I’ll see what I can do to get it done.
Have a great day, and may all your checksums verify correctly.