©Copyright 2004, NIIT and SLAC.
projects home maggie tools meetings presentations team docs
 

                                                                                                                       back

 

Thoughts for NIIT SLAC Collaboration

 

PingER

Analysis presentation of Internet trends

  • Possibly provide a Windows Web server with access to Excel to analyze and create plots of PingER data
    • Long term exponential fits, aggregation by region etc.
    • Number of hosts with loss qualities
    • Tie into web services/SQL access to data to make a much more flexible, distributed way of selecting analyzing displaying data
  • Do it with Perl scripts/GnuPlot flat files etc.

Set up a PingER archive analysis host in .pk

Based on experience with installing a PingER monitoring site develop a package to make installation much easier

Set up a web site for PingER

Extend PingER gathering to make it hourly so data is more real-time.

Extend plot-lat/long (see http://www.caida.org/tools/visualization/plot-latlong/index.xml) to provide support for different symbols

  • We  use this for displaying where PingER nodes are located, see http://www.slac.stanford.edu/~iepm/output.png
  • It would be useful to have such a package be able to create mouseover maps so could got to a URL or even just give the site name.

IEPM-BW

Maybe look into qiperf but need Web100 (using at NIIT may be a bit intense with todays NIIT network)

NDT (needs web100, active work is in progress in Internet2 so doubt this is a good idea at the moment)

  • Install an NDT server at NIIT
  • Command line interface
  • Make client lighter weight

Work on change detection, alerting – this is of immediate interest, is a tough problem, might be a good idea for a summer project then carry back to NIIT

  • Research, no guarantee of a product
  • On detecting a significant change, gather relevant information, e.g. RTT, traceroute, ping routers along route, get AS of suspicious points and get contacts. Look to see if can get more information, e.g. from other monitoring infrastructures such as PlanetLab, IEPM-BW, PingER which have “nearby” data.
  • Add to the detection the gathering of relevant information for the complete path.
  • Filter alerts (hysteresis) before reporting

 

 

 

IEPM-LITE

  • The richness of data collected at NIIT initially is limited since most connections go via London. So probably makes sense to focus on SLAC/CERN data for analysis
  • Given the low speed of the NIIT line it we do not want to impose much traffic so running the full iepm-bw with iperf etc. is probably a bad idea at first.
  • We are working on an IPEM-LITE that may be quite appropriate for NIIT
    • It does traceroutes, RTT & available bandwidth measurements
    • Measurement and analysis of  traceroutes from Pakistan might be very interesting, could show strange routes.
    • Looking for changes in abing performance  is low network impact, fold in diurnal and weekly variations
    • Should install at NIIT
      • Student works on install, nb. Not been installed elsewhere so needs work to make into installable package, this could be an NIIT project
  • Will need a fast traceroute as one scales to few hundred hosts (this is a nice stand alone project)
    • Write a new parallel traceroute, send probes (ICMP or UDP selectable) in parallel to different TTLs, provide many of standard traceroute features (-q number of probes, -w wait time, -I ICMP probe, -f first hop TTL, -F don’t fragment …, see the Linux traceroute man pages for ideas) plus add option to get AS information. Should not require root.

Student visit

  • We (SLAC) need named individual with a BS to whom we can issue an invitation so we can get started on getting the visa.
    • Focus on one student/project initially
    • Student visit will be to learn PingER and IEPM technologies so can take back to Pakistan