4 Replies Latest reply: Apr 23, 2012 12:28 AM by Chet Camlin RSS

Grouping and Dependencies

lthompson


I'm new to NPM, but have read the Admin guide covering Grouping and Alerting and the Technical Ref. on Using Groups and Dependencies and Understanding Alerts several times - but trying to apply what I have read to my particular network situation - I'm not sure of what would be the best practice to set up my groups and such.

 

I have a attached a rough diagram of a remote site of our network.  I understand that a group must be created for all the devices in Site A and then a dependency must be set up for the group for the point of connection (being either the connection from Corporate location or the local router).  However, in the Grouping Tech Ref. an alternate solution is given to group like items and create dependencies which flow from the WAN links down.  Our network is not set up as simplistic as the diagram shown within the example - so I am having a hard time wrapping my head around how I should set up grouping for this location.

 

My thought is to create a group (Group A) that would include the Accelerator and the Main Unit and then make it dependent on the Router.  Create another group (Group B) that would include all the device connect to the Main Unit (Controller 1, SW-144, SW-C1, SW-D1, SW-K1) and make it dependent on the Main Unit. Create a group (Group C) that includes SW-Maint and SW-I and make that group dependent on SW-144.  My question would then be how I would set up the switches that are chained off of SW-D1 and the Wireless Bridge and the devices behind the Bridge.  Didn't know if I would need to create another group which included all the devices after SW-D1 and make them dependent on SW-D1.

 

Didn't know if such a set up is too involved and would achieve the alerting that I would need.

 

Any help or suggestions would be greatly appreciated.  Or if there is any other documentation I may refer to, please let me know.  I have searched and read multiple posts concerning Grouping, Dependencies and Alerting - but it seems the more I read the more I'm confused.

 

Thanks in advanced.

 
  • Re: Grouping and Dependencies
    Chet Camlin

    You may be over-engineering the solution.  You should strive to keep your NMS system as simple as possible.  Just because you can create multiple groups with multiple subgroups does not mean you should.

     

    I assume the NPM server is back at corporate. 

     

    I’d determine what network resources are the most important in site A.  Create a static group for these resources so they always remain in that group no matter what is changed.  Set this group to worst status.  Set the appropriate alerts. If one device in this group goes down then the group goes RED and sends and alert.  The devices/resources in this group must all be operational in order for the particular service this group provides is available.  A number of companies use this method for monitoring email or critical applications.

     

    I’d also create a group containing the WAN router  (GP1) and another group for all other resources in site A (GP2).  Make GP2 dependent on group GP1 so you do not get hammered with node down alerts if the WAN link goes down.

     

    You do not need to group devices unless you can think of a reason why they should be grouped.   The most important rule of thumb is to keep the grouping as flat as possible.  Don’t do group>subgroup>subgroup>subgroup etc… 

    • Re: Grouping and Dependencies
      lthompson

      Chester:

       

      Thank you very much for the reply - the information was quite helpful.

       

      I have set up GP1 and GP2 as you have suggested.  However, my concerns are if Main device (member of GP2) goes down then I will get alerts on all the devices behind it, correct?  Or if one of the Wireless Bridges goes down, I will receive alerts on all the devices behind them - possibly making it more difficult to determine which device is truly down.

       

      Another question I have, possibly not on topic, but dealing with the alerting for GP2 - within the Tech Ref. "Using Groups and Dependencies" it states after creating the dependency for GP2 (the child) to GP1 (the parent), to create an alert for GP2 ensuring to be alerted when GP2 status is not equal to UP.  My question is if I would set up an alert to do that - checking the Group Name and checking if the Groups Status is not equal to UP - would I not get an alert if the Group Status went to UNREACHABLE if GP1 went down?  So would it also be necessary to check - and Group Status is not equal to UNREACHABLE also?

       

      Sorry to asks such basic questions - but trying to understand all the different elements that could be applied.

       

      Thank you, again!!!

      • Re: Grouping and Dependencies
        netlogix

        You don't want to send an alert email when something is unreachable or unmanaged.  You could have an alert that triggers on "down", and have that one's subject "Alert: " and a 2nd that is for not "up" and not "down" with "Info: ", then have a outlook rule that puts "info" ones in their own folder so you can look through them at your leisure, but address the important down "Alert: "s more quickly

        • Re: Grouping and Dependencies
          Chet Camlin

          Also, you need to evalute your network from two points of view.  Physical devices and Application services.  If you have a number of important physical devices connected to a single point of failure you would benifit from creating a group for these devices and make them dependent on the single point of failure.  From and application point of view, group all devices that the service depends on to function and set and alert based on worst case.  If one device in the group is down then the service is offline.