In the past the Exchange team and their MP developer peers already showed not being afraid of change or whole new approaches to monitoring Exchange server with SCOM, let’s take a look at the history of the MPs for monitoring the different versions of Exchange Server:
- Exchange Server 2003 MP
This MP had the (in)famous Self-Tuning-Threshold (STT) Monitors. In theory these STT Monitors are miracles of monitoring. In real life however they aren’t that good at all. The last iteration of this MP had many of these STT monitors replaced by the regular ones which reduced the noise significantly. This MP also used a tool, titled the Configuration Wizard for Exchange 2003 Management Pack. It helped people to get this MP configured in a fast way. However, this tool introduced new challenges as well. Overall this MP was noisy and had a high TCO because of the ongoing tuning.
- Exchange Server 2007 MP
This MP had no STT monitors what so ever. Also the separate Configuration Wizard used for the Exchange Server 2003 MP was dumped. Instead this MP was easily imported since all Monitors didn’t do anything. Simply because the related discoveries were turned of by default. Only one Discovery was running, the Discovery Helper. It showed people what Exchange 2007 servers were detected by this Discovery. When this was correct all other Discoveries could be turned on, one by one thus introducing a phased implementation of this MP. Many times the Discovery Helper identified servers as Exchange 2007 servers simply because the Exchange 2007 Server management tools were present. By tweaking this Discovery Helper process one could easily correct it, thus enabling correct monitoring of the Exchange Server 2007 environment. Overall this was a good MP and worked very well since it didn’t produced less noise and had a low TCO.
- Exchange Server 2010 MP
This MP is the most controversial MP the Exchange Server team introduced. It contained the Correlation Engine (CE) which was an additional process running on the RMS or OM12 MS server hosting the RMS Emulator Role. Again the theory behind the CE was very good, and still is. But in reality it introduced many unforeseen issues, like using some of the Custom Fields of the Alerts. Every Exchange Server 2010 Alert was processed by the CE which put some information into the Custom Fields of the Alerts shown in the SCOM Console. However, these same fields are used by Connectors and other similar processes which many time broke it. Another thing introduced were some whacky Data Sets which weren’t programmed very lean and mean. Because of it many of those queries contained in those Data Sets timed out, breaking the aggregation of raw data in the Data Warehouse, resulting in empty performance reports. This MP has seen many updates, some good and other outright bad. There was even an update which disabled mailboxes. So this MP has a whole history of its own and much of it isn’t that good at all.
No matter what one might think about it, it shows at least the Exchange Server team isn’t afraid of change and tries many new approaches when it comes down to the related Management Packs. And yes, many times I have cursed the Exchange Server 2003 MP and the Exchange Server 2010 MP. But at least these MPs go through a development cycle also after the first version has been released. Unfortunately the testing of those same MPs – before being released into the wild - didn’t prevent some serious issues hampering even more the already slightly negative experience index of the end users with this MP...
So now a new Exchange Server MP sees the light, the Exchange Server 2013 MP. And again this MP is totally different compared to its predecessor, because:
- The Correlation Engine is gone;
- The MP lacks the depth of its predecessor: it contains only some classes, and about 80 Monitors;
- No performance collection takes place what so ever;
- No Reports.
As the System Center Engineering Team states: ‘…each monitored Exchange server is responsible for monitoring its own health, and simply reports this via the Operations Manager agent. There is a little bit of roll-up going on, from Exchange server to Organization health. There are no special components running on the Operations Manager Management Servers…’
IMHO, this MP is just way too basic and misses out on collecting information required by many organizations. Also it plays down the role SCOM can play when monitoring a crucial service like e-mail. This MP reduces SCOM Agent to a proxy and skips all the intelligence present in any SCOM environment.
Yes, one could state the information required by those organizations is collected by Exchange Server 2013 itself. But why not expose that information to SCOM? Organizations want SCOM for many reasons, one of them being the SINGLE point where ALL information of many of their IT systems and services comes together. They don’t want anymore a huge collection of different points of information. A total overview is what they want and require.
Also the total lack of Reports isn’t a very good move either. IMHO, every MP should contain a set of Reports. Monitoring has grown up and has become far more then just looking at the current situation but also being able to report about how the situation has been in the past week/months and also being able to make future predictions.
So again the Exchange team and the related MP developers have chosen a course of action which I do not totally comprehend nor applaud to. I know my words sound harsh but I am just disappointed about the limited capabilities of this new Exchange Server 2013 MP. Hopefully future iterations of this MP will correct some of these ‘issues’ and missing (BASIC) features, like a mailbox count, mailbox size and so on…
The MP can be downloaded from here. The blog posting on the blog of the System Center Engineering Team about this new MP can be found here.
1 comment:
How many of the MVP's got to see this MP before it went public, and what feedback was given to them, if any, about this new MP? The reports in the Exchange 2010 MP are horrible, and the Exchange 2010 SLA one is flat out broken, never worked, isn't fixed. Reporting in SCOM (from the console) doesn't seem to add much value honestly, since many of them break over time, not sure that has changed with 2012 (doubt it). The fact no performance is captured is good and bad. Good because they often collect too much and it fills up your dw, BAD because if operators actually use the SCOM console to look at performance, then they are in a black hole now. What the Exchange TEAM needs to do now, is release addendum MP's that monitor specific components of Exchange and maybe even an addendum MP that does synthetic transactions. Build off the base mp. Knowing their history, and the fast dev cycles of MSFT technologies, I doubt this will ever happen. Seems to me they want to offer less, to make their cloud hosting solution seem like a better solution.
Post a Comment