Feature Request - Rule feature like in veeamone businessview

[vanhovejens]

Hi

I've installed the Veeam MP 6.0 a few days ago and linked it with our SCOM 2012 SP1 environment (It' really something magically happening )
Now after some time testing it I came to a point where I wanted to not get all the objects (datastores/resourcepools/vm's/vmfolders) in the SCOM monitoring.
This because some vm's are not cirtical/testing/dev/staging/... at the moment the only supported solution is to create a group in scom and make some dynamic rules to add the unnecessary objects to this group and disable the rules/monitors for this specific group with a bunch of overrides. (way to complex & causing way to much overhead)

Another unsupported solution was limiting the access of the readonly account on our VI but this also doens't seems to be the way to go.

Guess you know where my goal is?
There must be a much easier function in the nworks/veeammp back-ends.
If SCOM doesn't need all the object in the VI filter them not at the end but as close as possible to the VI to limit overhead & unnecessary resources get spoiled.
In VeeamONE there is a simple but very effective feature in the business view which solves this issue for our environment and for most of other environments i guess.
Just add some rules:
- do not monitor every vm which starts with "DEV"
- do not monitor apppools which equals to "Staging"
- do not monitor folder & descendant object "test vm's"
- ...

This is really a missing feature since it's already implemented in VeeamONE and we know you guys can get it done.
The collectors aren't lightweight machines when you monitor a decent VI environment, in our case we can filter out a few hundred vm's and a few datastores.
This really could safe us resources, effort and increase our flexibility which I guess you could only admit.

I've also discussed this with Mike Resseler from the Veeam team and he could only admit my concerns.

There are a few old & some very old discussions which also indicates this missing feature so maybe i'ts time to give it a try
Hope there will soon be an update with this feature included.

[sergey.g]

Hello,

Thank you for your waiting while we were investigating your request. Good question and good feature request by the way!

We are trying to understand what exactly can be done to create simple and elegant solution that will suit your needs. If there is a way to disable discovery for VMs with name which matches certain regular expression, will it work for you? We already have a way to disable VM discovery on a cluster by cluster level (you can move a cluster to a separate Monitoring Group and disable VM discovery in this MG).

Also we would appreciate if you can share with us more details about your infrastructure - how many collectors do you use and how many Clusters/Hosts/VMs do they monitor? We just wanted to make sure you are not experiencing any performance issues which force you to use more collectors. Maybe all of these additional collectors are not necessary or they can work more effectively if performance(or configuration) issues are fixed.

Thanks.

[vanhovejens]

Hi, thanks for your feedback

Disabling discovery which matches certain regular expression could be the fix for us. We use consistent naming for our staging/dev/testing vm's and by using the vm name we could filter for example all the "*-dev-*" vm's.
(we have been using veeamone and the feature in business view suits our needs).
It could always be nice to have more options for filter on folder/resource pool level etc. but starting with regular expressions will already be nice.

My concern is also that this won't be done by implementing discovery rules in SCOM itself because then the overhead isn't eliminated from the collector server. (filtering needs to be done at the lowest level possible)
We're not really having performance issues right now but limiting overhead should always be a concern.

Disabling cluster level or host level monitoring into other monitoring groups doens't suits our needs.
It's way to abstract and high level since we will have to create a separate testing/dev cluster with all the overhead etc. which we don't want/need

about our setup:
Currently we have one cluster with e few TB ram, on this cluster we've installed one veeam collector (to start with) and currently monitor about 150 vm's.
It's still a project that we will evaluate and after that we will see if we install some extra collectors and start monitoring a lot more vm's.

Thanks in advance to keep me up to date with this request and upcoming releases (is there already a roadmap for the new veeam MP? )

[sergey.g]

Hi,

Actually with such a solution we are applying this filtering exactly to the place where bottleneck occurs. All performance issues we saw were caused by the SCOM agent not being able to process configuration which our collector created. Also performing a filtering in a discovery script allows us to create MPs for different environments which may require such filtering or even add some necessary overrides directly into Veeam MP ISO.

Your enviroment should work perfectly, in our internal testing labs we were able to acheive 1000 VMs per Collector.

With respect to our roadmap - we are working right now to add support for new platforms, such as vSphere 5.5, SCOM 2012 R2, Windows 2012 R2 and to add some new features such as the one you've submitted.

Thanks.