Now that SCOM 2007 R2 RC is out…

In the last two weeks we’ve been pretty busy in evaluating RC of SCOM 2007 R2. btw this is one of the reasons we didn’t post anything. The evaluation is going great and I encourage anyone that’s serious about OpsMgr to evaluate R2. We joined the RPD program and validated the upgrade process of our entire deployment in our mock test lab, everything went smoothly and at this stage I see R2 as what OpsMgr should have been from the start but wasn’t. We tested the new Console (it moves at warp speed compared to SP1) and observed the full interoperability with sp1 agents, now we can’t help moving into production. And indeed we’re comfy we will move in a week or two, obviously with the full protection of the RDP program. To be clear I’m not saying you should move into production with RC Code. Only when we’ll move into production we can confirm our first impressions, I’ll keep you posted with any development.

Speaking of R2, you can find many references on what’s new and cool on the OpsMgr team blog, and insights in Marius blog, so I will just pick up my favorites (this is a shortlist):

  1. Overrides management. The long due integration of overrides in the Console will cut tuning time, it’s so usable that we plan to adjust overrides *after* the upgrade to R2. Initially we though to review our overrides (hundreds btw) before the actual deployment.
  2. Cook down now works for On Demand Detections, we will review our MP to add On Demand Detections and hopefully the new MPs will start to include this function. For more info on this kind of construct see my previous post “The downside of on demand detections”.
  3. Ability to scope the Object selection in report parameters. Now object selection can be scoped by class.
  4. image
  5. On the down side perf counter selection is still cumbersome and still we miss the ability to search on any field of the dialog instead of just on object name, but scoping is a step in the right direction (imo)
  6. Service Level Tracking is another addition we’re going to leverage, it’ still miss some whistles and bells but it’s usable and used with SLD (currently in beta) it will add that presentation layer that currently is missing. I’m expecting that the OpsMgr team will work further on this idea in future versions.
  7. On the MPs side, in the snmp probe provider has been added a WalkReturnMultipleItems that should make easier to monitor multiple instances in the same snmp tree (i.e. enumerating and returning data for multiple disk, int, etc.) Not yet tested. The whole system MPs have been revised with very interesting adds, I still haven’t had the time to test any of them (they’re not documented you need to dig inside the MPs), but they promise well.

So stay tuned as we move towards production prime time and we review our MPs, I guess we’ll find tons of useful and undocumented stuff.

btw Fabrizio and I will join MMS, maybe we’ll have a chance to get in touch in Vegas.

I wish everybody a Holy Easter

– Daniele

This posting is provided "AS IS" with no warranties, and confers no rights.

Technorati Tags: ,,
  1. #1 by Pete Zerger on August 3, 2009 - 3:21 pm

    It seems this is still not consistent in testing and not as reliable for sure. Marius comments that this is to be expected, which is something I did not understand previously.

  2. #2 by Raphael Burri on July 21, 2009 - 10:13 pm

    I must be doing something wrong here then. You’re saying that it should cook down, if the ProbeAction in the OnDemand circuit is called with equal parameters for all instances?
    In my experience unfortunately every single instance is still called independently although the ProbeAction’s parameters are equal.

    • #3 by Daniele Grandini on July 23, 2009 - 6:15 pm

      Very strange I don’t see such a behavior that was common before R2. Is your probe action script based? If so do you have any $ substitution inside the script? Such as $MPElement[Name=”System!System.Entity”]/DisplayName$?

      • #4 by Raphael Burri on July 23, 2009 - 9:16 pm

        Marius was kind enough to explain what is going on behind the scenes. It is not working because I have different instances aka. managed entities:

        If you are using a single data source per instance but drive various monitors, cook down in R2 will indeed work for OnDemand.

        By the way: The information you have on the blog that explain how you build up ypur composite modules so in a standartized way is very nice.

  3. #5 by Raphael Burri on July 21, 2009 - 10:00 pm


    Were you able to check if OnDemand detections are no longer being fired all at once when the agent configuration is reloaded? I just have seen that it still seems to happen with the RTM (Eval) bits. Would be interesting to see if you have managed to overcome that obstacle.


    • #6 by Daniele Grandini on July 21, 2009 - 10:05 pm

      Hi Raphael,
      in my experience they’re still fired but this time just once since cookdown is working, obviously if the on demand detection implements cookdown.


    • #7 by Daniele Grandini on July 28, 2009 - 11:10 am

      From what I understand from Marius post cookdown for ondemands for works per ME. I must confess I didn’t notice this behavior and I will soon step into my lab to check every single combination for scripts and oledb driven modules. For example I have a monitor with an ondemand that targets every single agent watcher (hundreds) and if this is the case I would have killed my RMS, but I must check. On the other hand, yes every time a monitr is initialized you’ll see two script instances, one for on demand detections and one for standard detections, you won’t have the second if it is time synced.
      Stay tune

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: