Skip to Content.
Sympa Menu

perfsonar-dev - Re: [pS-dev] New MA metrics proposal for discussion

Subject: perfsonar development work

List archive

Re: [pS-dev] New MA metrics proposal for discussion


Chronological Thread 
  • From: Nicolas Simar <>
  • To: Szymon Trocha <>
  • Cc: perfsonar-dev <>
  • Subject: Re: [pS-dev] New MA metrics proposal for discussion
  • Date: Fri, 25 May 2007 16:37:35 +0100



Szymon Trocha wrote:
Nicolas Simar wrote:
[...]

I would like now to make an ordered list of these new metrics:
1. L3 status
2. SNMP interface counters (in/out)
3. Ping results
4. Traceroute outputs.

I foresee at least 1 and 2 could be implemented within Y3. What is your view on the importance of such new additions?


Two angles to that:
- the utilisation
- the data available

Utilisation:
- During the last APM meeting, APMs have mentioned about their interests for interface input errors.


What do you mean here? I'm not sure I understood what you wanted to say writing "utilisation".

Oups, sorry, bad choice of words...
I wanted to say that there are two angles to that question: utilisation means the way people are interrested to use the data. The APMs mentioned that the input errors are quite important to them.

Data available:
- what data can be easily collected and fed to the MA? Are the NRENs already storing interface input errors so that we can make direct use of them?
- What tool can be used to populate those data?
- Are there NRENs ready to extend the data collection or their RRD tools?


For GEANT, we do not collect input errors, but we could do. We currently collect interface output drops per type of service (PiP, BE, LBE).


I assume that if there was the request to include this metric someone had to use it and that is why he proposed. For SNMP counters SNMP-MP seems to be the best candidate but in general MRTG is commonly used for storing counter values in RRD files. That means RRD MA should be extended with these metrics. Moreover even if somebody is not collectinf this type of data he would be able to do it in future. Anyway I would prefer to prioritize metrics which are _currently_ used somewhere or will be used in the nearest future in pS workplan (e.g. L3 status for CNM).

One of our main target is to support NOC and PERT. Best is to develop what they say. This might implies to prioritise the work on a SNMP MP and to make the RRD MA raedy for it.

Cheers,
--
Nicolas
______________________________________________________________________

Nicolas Simar
Network Engineer

DANTE - www.dante.net

Tel - BE: +32 (0) 4 366 93 49
Tel - UK: +44 (0)1223 371 300
Mobile: +44 (0) 7740 176 883

City House, 126-130 Hills Road
Cambridge CB2 1PQ
UK
_____________________________________________________________________







Archive powered by MHonArc 2.6.16.

Top of Page