DX NetOps

Hi Folks;

I was hoping to get this question out there and see if anyone has either developed something in house, or using another product to essentially identify what monitoring is enabled on specific devices.  For example, time to time we are asked from our customers "What monitoring is enabled / is monitoring enabled on CI abc123".  From speaking with members of my IM team, it seems that to answer that question, a significant amount of work, including checking the device in spectrum, then VAIM, then CA Performance Center, DB Insight, etc.

Has anyone developed any sort of mechanism to 'warehouse' this type of data.  Off the top of my head the only thing I can think of is a webservice that simply gets all models from spectrum, then queries all supporting tools through web services to see what policies/templates/etc are pushed out to each of those models.

Didn't want to re-invent the wheel here though if something is already available.

Thanks all!!

I am also looking forward to the discussion that this should start.  This is a question that we are asked from time to time and as you identified, we spend a large amount of time researching only to find the customer staring back at us like we are crazy.

In fact, today, I had one of our teams ask us simply:  "Do we have all our stuff in Spectrum and is it alerting properly?".  I loved that.  We finally agreed to take a mult-step approach together where we will retrieve the monitoring configs from Spectrum and they will identify what is "proper".

What we have setup today:

- We have a tool that another group in our company wrote that validates inventories in the tools.  What you can do is to take a list of devices, run it through this tool which then validates if the devices are properly setup for management (correct strings, correct access), you can ask it to verify if the devices are in the various tools, and you can ask it to model the devices in those tools, or remove them.  So inventory is not difficult for us.

- However, management configs is a completely different beast.  We have written a utility that will read the alertmaps, eventdisps, pcause and evformat files and document alarms for a given directory in the CsVendor or custom directories.  This information is put into an excel spreadsheet and allows someone from outside of our Spectrum world follow how an event may flow.

Looking towards the future:

- The next step that I would like to do is to somehow come up with a way to take a given device type.  Identify the mibs associated with the device type and then run the event/alarm documentation script for each of those mibs or groups of mibs to give someone a full picture of the device event processing in Spectrum.

-Then take that information and look at our SOI / Event Integration policies to show how it will flow into SOI or out to external tools.

If someone has other ideas, or has come up with a way to automate the documentation of device capabilities and management configurations, I would be very happy to learn about what you are doing.

Bill;

Thanks for the great response!  If you dont mind me picking your brain a bit, would you happen to have any knowledge of what the tool that one of the groups in your company developed does to identify the data that you mentioned?  I am wondering if they use web services.  I haven't done any deep dive into some of the available endpoints in the other CA tools (I primarily use SOI and PAM), but I am wondering if that is what is being done to `inventory` this data.

But I completely agree with you, this would be a VERY VERY popular tool especially for companies that monitor thousands upon thousands of devices and need this information readily available.

The tool that verifies the devices ability to be managed and inventories the tools was originally written in Python, with new extensions in Ruby.  It has a back end MySQL database where it stores the records of each of its runs for tracking purposes.  As you can imagine it tries to contact devices with SNMP, SSH, and other protocols to validate the device capabilities.  For the tools verifications, it is written in a modular way so that it utilizes whatever method that tool needs to communicate.  For example, for Spectrum it was using Modeling Gateway and they are moving over to REST calls.  One of our CMDBs utilizes an Oracle database that this tool queries for validation.  Other tools have their own separate interfaces.  So as you can tell, this was a massive effort to develop and requires constant updates as new device types are implemented and as management tools change.  But, it has a strong backing, so get's the support that it needs for updates.

M. Barnes,

Is all your monitoring within Spectrum, or do you use the VAIM/SysEdge for your servers? This split causes us headaches in trying to ensure that all drives/partitions are monitored, since VAIM doesn't make it easy. We have a complex, mostly manual process involving our asset management system compared to the contents of the files in the SM directories of our three VAIM domain servers, which are brought together manually, then combined into a csv by some clever scripting.

Closely related is the need for a good method to identify the SANM policies associated with the Global Collections, Topologies, Alarm types, and Models. There's no way within Spectrum to see what SANM policies apply to a given G.C. except by investigating each SANM policy individually. And we have cases where we used Models, such as for HTTP/DNS/TCP/ICMP tests.

There is the future of UIM and SOI, so perhaps there things will be easier.

Joe Poutre

BNP Paribas

That's our problem.  We have VAIM/SysEdge/CAPC used for monitoring.  I unfortunately don't manage those tools so I haven't yet identified the "how" of finding out what monitors exist on CIs, but from a preliminary thought process my only "where" is to place the data in our CMDB to warehouse the info since it is our Book of Record.

Would hate to lose traction of this discussion.

CA - Has anyone from the CA team considered this as a viable tool?  What are the limitations here of providing us essentially a manager of monitoring to show us what CIs have monitoring enabled and such?

You may have to document an "Idea" to get CA to look at this.  This has always been a challenge for us - how do we verify that the intended configuration and monitoring is in place without a lot of effort and time spent.  In the spirit of making the tools easy to administer, this sounds like no-brainer.

Done

https://communities.ca.com/ideas/235723001