NCM-REST ?

[ian robathan]

I was going to put this query as an idea but wondered if anyone in the community or CA itself can comment on this.

 

We are developing a new portal which will bring all our infrastructure monitoring needs under a SPOG (Single Pane of Glass) using software developed by ourselves.

 

Now without going to much into this Spectrum can not be the SPOG because of the business requirements but using REST we can obviously pull out what we need into the SPOG. However one area we are having to spend money on is an NCM tool. Basically Spectrum NCM is 'free' and OOTB but with no REST API For it in a SPOG you can upload config updates to it.

 

The idea is basically to trigger config updates from the SPOG to the NCM and for this we are testing out other vendor's NCM;'s which have the REST functionality and obviously will cost the company more to buy these.

 

So guess the question is in two parts

 

1) Has anyone had experience of trying this through Spectrum NCM, I guess not as no REST for it ?

2) Is CA planning any updates to Spectrum to allow for this and if so when, because if this is happening within 6 months or so then we can save some cash :-)

 

Cheers,

Ian

[ian robathan]

I am surprised no comments on this query, just bumping up for visibility in case some people missed it.

[Robert Kettles]

I think this would make for a good idea for the community to vote on.  As far as triggering a sync on demand, one thought is to look at triggering event 0x00821029 which is normally triggered by the device (Introduction - CA Spectrum - 10.2.1 - CA Technologies Documentation ):

 

When either of these traps are received, CA Spectrum generates event 0x00821029. This event then triggers Network Configuration Manager to perform a capture. If you want to trigger a capture for other supported devices, map more configuration change traps to that event. 

 

Perhaps a bit of a hack, so I would definitely recommend posting the idea.

[ian robathan]

Can I please flag this up again, we are still evaluating other NCM products and if REST is planned for NCM this would really be helpful

[Frank Tonjes]

The most I've done via REST is to view configs (running and start-up) but not done any pushing of configs yet as I have not needed to do this yet! Basically if there is a model in Spectrum for something - you can access attributes for that model.

 

I would imagine you could interface with Spectrum but it would depend on exactly how much you want to do. Spectrum doesn't do 'configuration management' as well as something like HPNA (I imagine you want to be able to create templates containing snippets of config for e.g. Cisco routers which based on some metrics builds the config) but this is a fairly bespoke area with many claiming to do it but not doing it well (We used NetMRI in my previous work and it's awful!).

 

Can you explain exactly what you mean by:

 

The idea is basically to trigger config updates from the SPOG to the NCM and for this we are testing out other vendor's NCM;'s which have the REST functionality and obviously will cost the company more to buy these.

 

You mention you are developing a new portal. I assume that if you can develop your own portal, this shouldn't be too much additional work - but have no idea what your requirements are so can't really comment.

 

One thing that takes a lot of effort is making it compliant to some standards (e.g. Security auditing, SOX, etc).

[ian robathan]

Thanks Frank for the comments.

 

A situation for example is this.

 

We ask the customer for ports they want us to monitor. When first added we will put a value against an attribute (say notes) on the interface which we then turn on monitoring for that port via the standard method and we will then update the interface alias to say it is being monitored.

 

What we would like to look at is where a customer will log onto a portal, change the port they want to be monitored (or add one) and this then flows through the workflow including changing the interface alias on the config then trigging a collection of the config, save config etc.

 

I was not aware you could actually show the config now through rest though and they may actually be a key thing, this is just an attribute value ?

[Frank Tonjes]

Yes, there are specific models containing the configs (HostConfiguration). This usually has a modelName of the device followed by the date/time the config was captured.

 

If you search for all modelTypes of HostConfiguration (0x820004) and view the attributes, you will see:

 

0x820073    StartupConfigText

 

and

 

0x82002b    HostConfigFile

 

These are the Startup and running config.

 

Oddly enough you can't work out the contents via vnmsh (at least I haven't found out how to as I think they are encoded). If you use REST and query the value of the attributes you get the ASCII character codes (separated by periods (making up the config). You can just iterate through the string and convert the characters e.g. print chr($int). There are some non-displayable characters like newline and others so you will need to handle this as well.

 

For configuration of ports, we actually do it the other way around. We require the network teams to add a descriptions and we have scripts which look for this at deploy time and then enable/disable interfaces. It's also easy for other tools like CA PM and eHealth to pick up.

 

Typical example is to use something like [PRIMARY] and/or [SECONDARY] in the description and then just update the attributes on the port to enable/disable the monitoring.

 

This way the responsibility lies with the networking team. Alternatively if you have a CMDB and this info is in there you can just use that. (not sure how you decide what ports get monitored).

 

Also see: https://communities.ca.com/message/241893168#comment-241893168

[ian robathan]

Thanks Frank, that is something to think about.

 

The method you mentioned about interface alias is one I have used elsewhere but on the current job they are looking to try something different and let a customer be flexible, ongoing debates about that :-)

 

The problem being is that we never have had a alias convention and would need to backtrack on thousands of ports to do this.

[Frank Tonjes]

Yes we have that same issue and have finally worked out the new standards and are busy rolling them out. We also looked at some way the customers can enable/disable monitoring but in order to be CAS(T) compliant we can't let customers influence configuration directly and are getting them to raise tickets instead to get us to change the configs!

 

Let me know if I can help with anything else and good luck. Hope you manage to get it sorted!