perfsonar development work

[Nicolas Simar <>]

Joe Metzger wrote:
> Hi,
> I am trying to identify appropriate tools for the LHCOPN community to
> deploy to monitor latency and congestion on their links using the
> perfSONAR framework.  I am concern because they need to
> start moving forward with monitoring now. And I am not sure all
> the necessary perfSONAR components exist yet.
>
>
> The expected deployment would look something like:
> 1. Each site deploys a one-way delay measurement point (MP).

What tools have we got there:
- Hades: linked with HW and the information are stored in a central DB 
system for different locations.
- CL-MP using OWAMP.

> 2. Each site deploys a measurement archive (MA).

A measurement archive for the one-way data or for the link utilisation?

> 3. All the MA's and MP's register with a common LS, (or a
>    set of cooperating LS's in a multi-LS environment.)

Should be the case for any pS2.0 release.

> 4. Each site configures their MP to run continuous
>    tests to their neighbors of interest, and publish
>    the results in their MA.

There are schedulers for some tools and a way of pushing some data with 
the Hades system. I do not know for the CL-MP and owamp.

> 5. A GUI allows end users to visualize the data.

Hades has got some own (installed on the same server as the data 
archive, but not retrieving data with personar)
pSUI and CNM allows the display delay information.

> 6. Some sites with NOCs, deploy an alarm tool that regularly
>    fetches the data for a given subset of links from the
>    MA's and raises alarms if the performance is out of spec:
>    IE no data for a recent period, or the latency exceeds
>    some threshold criteria.

I know DFN is working on some alarm system (delay). But it is still in a 
development stage. If you have recommendation on what type of alarm you 
would like to see, this would be a good input to the process.

EGEE has also been working on some tools. I do not know their status.

How do you intend to proceed?

Nicolas

> I am aware of efforts to address some of these issues, but I am
> not sure when it will all come together.
>
> I expect the current RRDMA and/or the SQLMA and the LS will meet the needs.
>
> I am not as clear about the MP, or the component that will
> cause the MP to make continuous tests. Who is working on these?
>
> Will the perfSONARUI work to visualize the data?
>
> Is anybody outside of SLAC working on data analysis and automatic
> alarm generation?
>
> --Joe

--
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
_____________________________________________________________________