Tuesday, November 25, 2014

You’re In Charge For The Next Gen MPs. What Would You Do?

Yes! That’s right!

Suppose YOU’re the boss for the team building the next generation of Microsoft Management Packs. And there are NO limits. No budget constraints. And limitless resources are available. There is only one thing to do:

Build the next generation of Microsoft MPs which will blow everyone of their feet

What would you do different? What would you scrap? What would you keep and what would you add? Also think about presentation (dashboarding, reports and so on). What kind of functionality are you missing in todays Microsoft MPs which should be present in the next generation?

I know. A situation like this won’t happen. BUT…

Suppose I’ve got contacts. And they asked me to provide them with information about what the next gen of Microsoft MPs should look like. What those MPs should do and shouldn’t do. How those MPs should function. What these MPs should fix in todays MPs.

And no, this is NO fantasy. As a matter of a fact I’ve been asked such questions quite recently. And I’ve answered them. But you know, that’s just me. I know many people who are working knee deep in the boiler room of SCOM, AKA ‘the trenches’. So why not ask them as well in order to get a WHOLE list of items my contacts can work with? Wouldn’t that be awesome?!

That’s why I am asking YOU to tune in and leave your comments on this posting.

Those comments will be forwarded to the right people who are about to start soon on the next generation of Microsoft MPs. I know for sure they’ll do their utmost best to incorporate as much as possible of those highly valued comments. They’re eager to hear from YOU!

So speak NOW and changes are your highly valued comments will be incorporated
into the next gen of Microsoft MPs!!!


Cameron Fuller said...

1) Cloud monitoring is priority #1. If Microsoft is going to bet the company on the Cloud, they need to bet their management packs here as well. Priorities include:
a. Fix the Office 365 management pack – it is unusable due to the alert re-opening issue (and generates more alerts than all MP’s combined in my environments).
b. Provide an Intune management pack – all SaaS and PaaS must have solid management packs from Microsoft. The lack of an Intune MP is just bizarre.
c. The existing Azure management pack – needs to be truly deep. It needs to be able to see everything (other than IaaS) which should be available at the top level of this stack. Common and custom performance metrics need to be included.
d. Integrate the Azure Operations Insights into the Operations Manager console natively through a management pack.
e. Provide monitoring for Azure Operations Insights in Operations Manager.
2) Tier 1 workloads: Active Directory, Exchange, Windows Server, SQL. Any enhancements in any of these technologies needs to be effectively monitored in Operations Manager.
a. Specifically on the Windows Server MP - include metrics for IOPS a drive is capable of not just how much traffic it’s presenting. This is key when capacity planning for drive usage.
b. SQL MP’s – See (http://blogs.catapultsystems.com/cfuller/archive/2014/10/29/recent-updates-to-research-this-sysctr-scom.aspx) for details, but the alerts listed there need to be re-designed as warning level not as critical level on the health model.
c. Dashboard management packs: Additions of dashboard management packs for all Tier 1 workloads.
3) Re-write the Group Policy management pack: The usage of manual reset on almost all key monitors renders this management pack effectively useless.
4) Any non-native management packs: Anything which is still out there which was converted from MOM 2005 just needs to be removed or preferably re-written into a native management pack.

Wilson W. said...

My suggestions are more generalities than specifics....

I suppose, better MP documentation that shows exactly what has changed from one version of an MP to another. (I'm talking detailed documentation that someone could submit to a change management team for approval)

No more of these inexplicable changes such as the changes that were made to the Base Windows OS management pack for disk space monitoring. The disk space monitors used to report how much free disk space was left. Then they suddenly changed it so that this information was abstracted. Then they put that feature back IN again. None of that was documented very well.

No more MP's that bring your SCOM environment down to its knee's (Exchange2010 MP and it's correlation engine).

More out-of-the-box dashboards that take advantage of the new dashboard widgets that are available in R2.

Perhaps a way to link monitors and their corresponding rules together?

Unknown said...

01. +1 for 1)d in Cameron's comment.
02. Being able to generate SCOM alerts based on Saved Searches from OpInsights like this example based on event log searches: http://www.muscetta.com/2014/11/22/capturing-your-knowledgeintelligence-should-be-simple/ but being able to bring the result back to SCOM as alerts.
03. ConfigMgr management pack. If SCOM's native text based log file monitoring capacity can be improved, and maybe ConfigMgr management pack can be extended to monitor critical issues from the hundreds of log files within ConfigMgr. I see ConfigMgr people spend a LOT of time everyday reading different log files on different site systems and often critical entries have already been overwritten by the time they get to it. I believe SCOM should be doing this for them!
03. provide a way to remove pictures from the topology dashboard widgets (http://blogs.catapultsystems.com/cfuller/archive/2014/05/27/quicktricks-removing-images-from-the-topology-and-image-widgets.aspx)

Anonymous said...

Dashboards for the basic infrastructure components, such as Windows, AD, DNS, would be very nice. Some more useful reports (such as free disk space) would be useful to. This also applies to the Exchange Management pack. Why can’t I see basic stuff like mailbox and database sizes?
Other things I miss is good documentation. Some MP’s (like SQL) does a pretty good job but the Exchange MP is still lacking detailed information. I find it especially hard to find out what all the rules and monitors are doing (the actual script). Doesn’t understand it anyway why there is no easy way to find the content of a script.
Some dashboard improvements would be welcome as well. Especially creating dashboards is dead slow and the consoles crashes frequently.

Simom Skinner [MVP] said...

vNext did a MIBs to MP cloud tool, I would like to see a Cloud based tool that would verify bespoke MPs, in additional an auditing capability for versioning. MPs are and will continue to be a key part of the success of System Center (SCOM & SCSM).
Companies always have apps that don’t and may never have MP’s unless there are built internally, VSE & MP Author are tools we can use but both come with their frustrations. VSE in my mind is the better way forward so we can use TFS for storage and sharing.
OpsLogix have a MP authoring Fridays run by Oskar Landman, there is so much you can learn from these guys to create that ‘better’ MP.
Companies need better processes internally also to check the structure of the MP before they go in to production.
However, on the whole this is still a current subject. Well done Marnix for raising it….again!

John Bradshaw said...

I would also add, hoping not to be rude, but direct in some comments:

1. Reporting. This has been a looong bugbear that Cloud based monitoring rivals to SCOM win hands-down! We need one-click reports for the common server attributes. cpu, mem, etc. Has anyone thought of adding Wizards to guide through the creation of a new report? All we get is a screen that is blank and Add Group or Add Object and we have to spend hours figuring out what SCOM requires to be inputted. And then a puny graph comes back that doesn't even have labelled axes!! What is needed is to be able to get pretty reports easily for management.
2. Easy SNMP device monitoring. It really is painful in its existing state. E.g. Import mib file and voila!
3. URL monitoring. (We gave up on SCOM and went to an external 3rd party which is one-click on everything to do with URL monitoring. Reliable, can monitor from 30 different countries...all available on my fablet...SCOM is way behind for ease of use here).
4. This may raise some feathers, but in all companies I have been involved in Unix/Linux monitoring is always conducted by non-Microsoft people. They won't let SCOM within a bull’s roar of their precious systems. Why is SCOM even bothering to get in to this field? Better to focus on getting Windows right.
5. As for Publishing to SharePoint….well it would be nice, but life is too short. If I have the right credentials, it should be made easy via gui.
6. I have had Microsoft people sit with me, at my desk, trying to get a new dashboard or two working and they have given up. Surely that says something. The widgets idea is too time consuming to get the right data displayed.

Sulick said...

some more:
1) Relative dates picker in every report with start\end date parameters.
2) Possibility to choose avg\max\min SLA value in SLA reports for a group of objects. Now it shows minimum(worst) only. Can be done manually, but... :)
3)SNMP- some way to import mibs. It's a little bit of pain to build a new MP for every unsupported device.
4) Make discoveries that rely on registry work with Windows Server 2012, etc (MBAM for example)

Bjørn-Erik Løken said...

Maybe a bit late, but here it goes.

+1 to all of Cameron's comment. I need to point out that Camerons nr. 2: the AD MP is in need of a complete re-write. It’s too complicated, too rule based and too "untrustworthy".

Always utilize Resource pools whenever possible, use the possibility to configure proxy if it's Azure or Office 365. Instead of using a watcher node (Windows Azure SQL MP).

Lync Server MP: active monitoring, synthetic transaction main script, has a problem whit Powershell execution policy. They use –bypass, but that will not work if you have the group policy set to “unrestricted”.
Offer an “Azure Watcher Node for Lync” – Build it in as an option in GSM and link it in to the active (synthetic) monitoring MP.

Windows Server Remote Access 2012 MP – Adds two views to the Root Monitoring Tree: “RemoteAccessSite Alert”, and “RemoteAccessSite State” I don’t like that, it’s enough that they can be accessed from the MPs view folder.

Cluster MP – I think the rules could be “upgraded” to monitors if possible. I’m not an expert on clusters, but it would be nice. We sometimes have many “unable to access witness resource”, “cluster has exceeded failover threshold “ and “cluster service is shutting down” due to short network blips. Auto-close if possible please.

Keithk said...

SQL MP - The ability to measure performance at the query level. This is the number one reason why our database team chooses to use Foglight.

SSRS MP - The ability to measure performance at the report level.

Exchange MP- More granular visibility and reporting around mail flow.

For all management packs to have the following:

1) Drill down in context dashboards -
Out of the box dashboards and/or widgets to allow the creation of dashboards that allows users to select objects and be able to "drill down" into "in context" dashboards. This is seen in Veeam with their analysis dashboards.

2) Better summary dashboards -
Out of the box summary dashboards that cover many areas of health and performance that summarize a performance graph, numerical value, and color status in a small tile. This is best seen in the SQL mp. Allows for quick assessment of health and performance and a very efficient use of real estate.

3) Personalization thresholds in dashboards -
The ability for users to personalize their own critical threshold color values to correlate with performance metrics in a performance chart. This is seen in the Veeam 7 mp with their traffic light capability.

4) Top N dashboards -
Out of the box top N dashboards like seen in the Veeam 7 MP.

5) Heatmaps -
Out of the box Heatmaps like seen in the Veeam 7 MP.

6) Integration with APM -
SQL for APM perhaps to measure query execution.

7) Spinning Wheels -
Don't laugh, but my customers were like hypnotized by the spinning wheel from vFoglight. Perhaps animated arrows moving back and forth with green, yellow and red. The point is more interesting ways out of the box to represent data beyond a line graph.

Unknown said...

Big +1 on both Camerons, Simons and Johns comments, especially on Reporting and Management Packs. A lot are too complex and badly written.

My own comments:
1)i'd like some better Powershell integration when it comes to monitors. Create a monitor based on Powershell from the Console - Wei Lim has created a management pack, but that should be built in. Also, the possibility to create a performance rule from a PS script.

2) Windows Service template: parameter to control downtime, so one can choose that a service can be down for 5 minutes before alerting.

3) Exchange 2013 mp. This one is... bad.

4) VSAE: better documentation/guidance of the different areas for better understanding.

5) Monitor/rule/discovery debugging: debug what a monitor actually does/does not

John Bradshaw said...

I was working with a Microsoft SharePoint sme yesterday and we were trying to get some in-depth monitoring going for URLs. We both concluded that we like our Dentist better than SCOM on this point.

It would be great if some of the helpful people at Microsoft would put up some real life video "How-To's" that are current for 2012R2.