perfsonar development work

[Yee-Ting Li <>]

Hi,

Sorry for not replying earlier (christmas/new years hols and all).

The 'gmaps' suite isn't really a 'weathermap' per se; although one of  
the goals is to be able to quickly and easily discover and present  
network information from a number of different perfsonar sources.  
Currently it supports the snmp ma and the rrd ma perfsonar interfaces.  
it also supports display of pinger information on latency and jitter  
measurements.

A example instance (i'm not sure if all the links work as it was set  
up for the SC2007) is at

http://packrat.internet2.edu:8008/

i don't believe it will work under IE6/7, but should be fine under  
firefox.

basically there are a couple of features that i would like to add:

- colouring/symbol of nodes to represent state
- addition of coloured links between nodes (with dynamic bezier curves/ 
lines)

in turn, the above should be able to quickly show link status and link  
quality.

of course, there is need for more useful debugging information.

the main issues for me are the need to cross correlate node/link  
information across different sources. i currently do this is pseudo  
'urn' descriptions (as no services except for some of aaron's stuff)  
supports it.

Anyway, i would be more than happy if i could get 'gmaps' to present  
any and all perfsonar network performance information, and more than  
willing to work closely with you guys to ensure it so if you think  
this may be useful to you.


Yee.

On 4 Jan 2008, at 14:18, Roman Lapacz wrote:

> Andras Kovacs wrote:
> > Hi Roman,
> Hi
> > The testing is only about to test the signalling, not the traffic.  
> > If there is packet loss inbetween, it is the task of PERT or NREN  
> > or/and GN2 NOCs to solve the problems.
> >
> > Here, we would like to only check that the international  
> > infrastructure allowing countries to peer works or not.
>
> I see your point but it could be very useful for troubleshooting to  
> monitor both levels: 1) signaling and 2) network transmission (even  
> using simple ping) in the same infrastructure. Doing regular test  
> you would have quite complete view (signalling information and base  
> traffic statistics).
>
> > > First we need to investigate and write down the names of metrics  
> > > to be monitored. I think in this kind of monitoring important are  
> > > delay (owd), jitter and packet drops (of course there might be  
> > > more).
> > In the very first version (until the end of GN2 project), we only  
> > plan to measure up/down, it is enough to cover our needs.
> >
> > > I guess that Andras wants to focus on passive monitoring  
> > > (monitoring the traffic during VC transmission). A test tool like  
> > > BWCTL could be used as well but test traffic should have  
> > > characteristics similar to real VC transmissions.
> > This is not, right, see the aboves.
> >
> > > > c) How will they access the information? (what visualisation  
> > > > tool, you hinted one below) How will it be the data represented  
> > > > to them?
> > > > You mention a clickable weathermap. Can you provide a little bit  
> > > > more information about this.
> > > > (please see a status page: http://perfsonar.acad.bg/status/ and  
> > > > how we might be building another similar page for the network  
> > > > services at a later stage: http://wiki.geant2.net/bin/view/SA3/SaThreeAppNrenStatusPage)
> > >
> > > The application made by Yee could be also considered (it's based  
> > > on Google maps). It looks really nice and I remember it had very  
> > > (!) positive opinions during last I2 conference.
> > Is there an example to see this application? We would like to use  
> > geographical representation for other purposes, we might want to  
> > integrate monitoring into this.
>
> Knowing more about your plans I don't think this app is for you.  
> Maybe Yee or Nicolas could say more.
>
> Roman