You are viewing an old version of this page. View the current version.

Compare with Current View Page History

« Previous Version 4 Next »

Performance WG meeting

10/4/11

http://events.internet2.edu/2011/fall-mm/agenda.cfm?go=session&id=10002087&event=1148

1) Celeste opened the meeting with a presentation on perfSONAR deployment. Specifically mentioning the PACWAVE survey and encouraging everyone do fill it out. It should only take a couple minutes to do so.

2) Joe St Sauver gave a presentation on HTTP. The general idea was that because of the increased number and types of attacks sites may need to move to HTTPS for any content delivered from the site. For this group the question is will doing so impact the performance of the the site?

Joe's Slides: http://www.internet2.edu/presentations/fall11/20111004-stsauver-hsts-performance.pdf

3) Jeff gave a presentation on the performance service that Internet2 is considering. As part of the discussion, Joe Metzger pointed out that the DICE collaboration is really about making sure that PS and MD are fully compatible. Joe also suggested that the main push is to get the infrastructure in place to solve the hard problems. The easy ones we can already largely take care of. Joe St S. asked what had happened to netflow. internet2.edu.  He also pointed out that the AMP program was very useful and provided good information about routing and traffic.

In response to a question about what would be useful, it was suggested that a weather map sort of tool would be useful to determine:

1) Just how long it should take to move a dataset from A to B.
2) Reliable empirical data about moving data
3) Statistics would also be useful

The idea is to look for something predictive rather then historical.

Someone suggested wanting something like a display that, in real time, showed the use/utilization of various paths to help allocate network resources.

There was also a mention of taking into consideration various regulatory issues to provide data about performance. The FCC or others might be relevant here.

  • No labels