All Interfaces are not Equal

Donald_Francis

Something I always struggle with and constantly have to workaround is the fact that by default all interfaces are equal.  What I mean by this is there are interfaces I monitor primarily for stats only and only care about when they toss mass errors or something to the like.  But then there are others I need to know if they are up/down and if they have any errors at all.

So for the most part nodes are nodes and while they do get chopped up into families or groups (IE network vs systems) they are essentially reacted to the same in my neck of the woods within those large groups.

But again with interfaces I have these different sensitivity levels where some I will only pay attention to after a million errors and some I need to know essentially if there are any or they go down.

Now obviously this can all be done with custom properties and that is fine I am just putting forth the thought process that maybe interfaces have a built in way to designate the common priority levels that seem to innately be there.

Just for example this is the interface groupings I use in my head, custom properties, filters or what have you.

 

-Router WAN (IE the actual WAN connection on a router not the LAN connections on it)

-Switch WAN (MetroE interfaces which on a switch end up being just another LAN interface physically but one with a very special duty and priority).

-Switch Host Interfaces

-Switch Uplink Interfaces

-Switch Server Host Interfaces

 

Furthermore it would be awesome to maybe there be say,  just built in check-boxes or radio buttons to use.  And on top of that maybe some just basic logic auto selects what type of interface it is.  This could be based on say interface description or name.  IE anything containing the words "Host or workstation" in the name could be tagged as host ports.  Anything with "Server" would be a server host port and uplinks could be recognized by simply "uplink" or something similar.

mdriskell

Donald, 

I do this using custom properties because I agree they are not all created equally.  For instance we use several multilink circuits in our environment so I tagged the aggregates as WAN_Interface and then filtered some of our views.  The reason for this is if I had 3 T1s on an router that made up 1 multilink interface if they were overutilized I would end up with 4 interfaces in alert, so I filtered on the WAN_Interface condition.

Furthermore it would be awesome to maybe there be say,  just built in check-boxes or radio buttons to use.  And on top of that maybe some just basic logic auto selects what type of interface it is.  This could be based on say interface description or name.  IE anything containing the words "Host or workstation" in the name could be tagged as host ports.  Anything with "Server" would be a server host port and uplinks could be recognized by simply "uplink" or something similar.

As for your automated selection I don't know about that but one way you could accomplish that would be with a SQL query.  You could do this in two ways.  First you would need to create the custom properties you mention, Router, uplink, etc.  Then you would need to write a report looking for all interface captions of "Uplink" that didn't have the Uplink custom property set as true.  That would allow you to go and target which interfaces needed correcting.

If you are brave and comfortable in SQL you could write a query that could automate this for you (I have done similar tasks in the past) but I am always wary of doing it in this fashion.

Once you have your conditions set then of course you can do whatever filtering or separate alerting you require.