Using Virtual Server R2 To Resolve Microsoft Operations Manager 2005 Issues

I have a new project under way. It was born out of several ideas simultaneously. First, I needed a test environment for work with MOM 2005, and the transition to SCOM 2007 (still in beta). Second, I found out that Insight allows business customers to now host web and email servers (you have to sign a business contract). This seemed like great timing for these two things to happen, because I could now host my various sites, email, and have admin control without having to pay for a co-located server. So, I setup Virtual Server 2005 R2 on a Dell 1600SC Server (Xeon 2.4Ghz, 2GB of DDR Ram, and four 36GB U320 Hot Swap SCSI drives). I built four virtual machines as you can see in the image. The names should be self explanatory.

Most recently, I upgraded to SQL 2005 so that I could more easily migrate to System Center Operations Manager 2007. The current plan is to change the setup for SCOM2007 within the next week.

I just came back from Tech Ed 2007, where I had a chance to meet a program manager for the MOM product. It was an informative time, and he highlighted some things that they are working on (which may or may not show up in the 2007 release). One of the biggest things that interested me was the addition of a more advanced “Health” association. Basically, in order to determine how healthy a certain database is, you need to associate the items that have an affect on it (the database runs in an instance of SQL, which runs on Windows server, which has hardware, etc..). The neat thing about providing association is that it brings about the ability to do improved alert suppression (which actually was my primary interest).

As an example of the current problem, here is a scenario: You have a complete Windows network, with some Exchange servers to host your collaborative communications. To provide an aspect of scale, let’s say there are 16 AD Servers that act as Global Catalogs, and you have 100 Exchange servers. You introduce MOM monitoring into the environment, which is a very proactive solution for knowing the health of your systems. Each server has a MOM agent running, which reports periodically to the MOM servers (based on scripts or events). One feature, called AD client side monitoring, allows you to monitor the health of Active Directory from a “client” perspective. “Client” in this case refers to other servers such as Exchange, SQL, etc. One aspect of client side monitoring is the “AD Client GC Availability” script, which runs every 5 minutes to make sure the client servers can see a GC.

You decide to reboot several of your GC’s do to updates. You do it during an approved maintenance window, and you even put the machines in “maintenance mode” for MOM so that it doesn’t generate excessive alerts from intentional reboots. The Servers start rebooting, and the GC Availability happens to kick off on some of the Exchange servers. All of the sudden you have are filling MOM with alerts about the GC’s being unavailable, yet you already put them in maintenance mode. This is because those scripts are looking at things from the client perspective, and the client is not in a maintenance mode.

This is a difficulty currently faced in MOM 2005, and it may very well be in SCOM 2007. It was nice to know that they are working through it, and trying to come up with good options for dealing with this problem. It is certainly not a “one size fits all” issue, so there is some difficulty in how a solution should be provided.

more updates to come as i put on SCOM 2007.....