Latency graphing?

[paul.baguley]

Hi all,

 

We are using an IM2 suite with Spectrum, Performance Center, Data Aggregator and some additional data sources such as UCM, NFA, ADA...

I am looking for the assumed capability to report Latency on discovered devices (network models primarily) but don't seem to be able to see anything reagrding this metric within the product OOB...surely this is covered?

 

Could someone point me in the right direction please...graphed out ping response times for our WAN map is the desired end result.

Cheers!!

[Anon Anon]

I did this with some servers with the Dynamic Trend View

View Type was MultiTrend

Metric Family was Reachability

Metric Value was Average Response - Average

It came out looking like the screenshot, is that about what you want ?

[Baljit Bansal]

can someone advise how i get this for interfaces and not device?? really looking for tabular form
 

[Anon Anon]

baljit-bansal:

can someone advise how i get this for interfaces and not device?? really looking for tabular form
 

Since you can't really ping interfaces, that's going to be a bit hard...

[Baljit Bansal]

how does eHealth do it then? you are able to report on latency on interfaces through a topn report, im looking for the samething but in data aggrgator??
 

[Stuart Weenig]

I don't know how eHealth does it, but in order to do it right, I would
expect to discover all IP addresses on a system then ping each of those IP
addresses. If that's how eHealth does it, PM2.0 doesn't do that today.
You'd have to put in an idea.

[Anon Anon]

I would assume that eHealth just gets the latency for the IP address of the interface, and not the interface itself...

Perhaps an example of exactly how the server is configured and what you need would help ?

[Anon Anon]

This is a problem very bad where work, every needs reports with latency or average response a before they could realize in Ehealth.

Any find the solution for this error of design? Maybe from vendor certification or adding a new metric families, but is a large development.

[Renato Orgito]

I'm curious: why do you need latency for each Interface? I don't think there will be any significant difference in latency from different interfaces in the same device.

[Stuart Weenig]

Really what he's probably looking for is path latency. One router with two interfaces on two WAN links will have different, significant latencies. Knowing the latency of each compared to the latency to the device itself (path to the loopback) will give a good indication of performance of each distinct path.

[Renato Orgito]

Makes perfect sense! The way I can think of doing it is using specific routing and IPSLA.

[Stuart Weenig]

Yes, this is best done with a full mesh of IPSLA tests. 

[Anon Anon]

Yes, the best practic is use IPSLA, but in my work this task are performed for different groups, my group only administer this tool and others (SCOM, ITNM, Omnibus etc) but not configured any cisco and other similar device. Is all a problem request ipsla, too much bureaucracy, in addition Ehealth show this variable for each interface, CA PC at the very least should the same.

[Stuart Weenig]

From the beginning, we've asked for PM2.0 to have all the features of the products it is supposed to eclipse (eHealth, NV, Spectrum, NFA). While it looks like the Spectrum and NFA functionality will never be built in due to a change in targeted audience for PM2.0 (mega-service provider instead of enterprise), many of the features of eHealth and NV have not yet been built into PM2.0 either. It's possible these features may never get built in due to the aforementioned targeted audience change.

 

All of that is to ask you to clarify your request. Currently PM2.0 doesn't track per-interface latency. Obviously it also doesn't report on it. Asking for a feature to be built into CAPC is not the right way to go. What you want is for the metric to be tracked in PM2.0. This may be possible with a custom vendor certification if eHealth is using simple SNMP to track this data (this is something any customer can do if he knows how to find the data in the MIB). If eHealth is using a special engine to track ICMP based metrics like latency (most NMS systems do), then a special engine will have to be built into PM2.0 for this (which will have to wait for CA product management to deem it important enough to the mega-carriers; see first paragraph of this post).