IEPM

Status Report for SLAC/DOE IEPM Project

For December 2001

SLAC Home Page
Work completed or in progress since the last report.

Work Related to Analysis and Development

IPEX

Warren worked with Allen Mcintosh of Telcordia to configure the new IPEX passive monitoring box for SLAC.

Certificate Authentication

Warren wrote a guide to configuring certificate authentication.

IPv6

Warren worked with Bob Fink and Perry Metzger to supply Ipv6/6bone performance data. Informed Bob that the ipv6 monitoring is no longer active and discussed Internet2 support for it.

Tools

Grid-Monitoring

Warren documented the network monitoring in terms of usage scenarios for the joint PPDG/Griphyn/IVDGL monitoring group.

Web100

The SLAC Web100 evaluation has been split to 2 machines. One will be stablised to allow testing and the other will be used for development.

Missing Data and New Sites

Web Stats

Warren resurrected the web-log analysis.

WAN Utilization

Warren worked with Connie Logg to reformat historical data of the utilization of SLACs WAN connections.

New Sites

A new monitoring site at Trieste has been added.

PPDG/Griphyn Monitoring Group

Les wrote a proposal for closer cooperation between the IEPM project and the PPDG. This was sent to the PPDG executive committee and following a discussion among the executive committee they enthusiastically accepted the proposal. Following this a phone meeting was held between the PPDG executive and the SLAC & FNAL IEPM folks, and the collaboration was inaugurated. Les wrote a quarterly report of the IEPM activities for the PPDG, see Report on individual activity to the PPDG for the quarter ending December, 2001.

Web Stats

Warren resurrected the web-log analysis.

WAN Utilization

Warren worked with Connie Logg to reformat historical data of the utilization of SLACs WAN connections.

Active End-to-end Throughput Measurements (IEPM-BW)

Les worked with: Les, Connie, Fabrizio, Jerrod and Manish rewrote and extended extend IEPM-BW from the quick & dirty code thrown together for SC2001, and to improve reliability, configurability and testability, by one by one addressing problem areas for each host. Les updated the sc2001-bwc email list and got Teresa to change the name to iepm-bw. Les investigated why pipechar was failing on pharlap, it looked like a hung port. It seems to clear itself up. Gary thinks it maybe that the ports are in TCP_WAIT, which eventually times out. Les and Fabrizio started to analyze new data to compare bbcp versus iperf throughputs, vs pipechar. Looks like bbcp~0.67*iperf, and for most cases, as long as the disk throughput is under 80Mbits/s bbcp disk to disk ~ bbcp mem to mem.

Les worked with Andy to get new bbcp with compression working, and with Andy and Brian Tierney of CERN to get Netlogger into bbcp.

Work related to Publicity, Outreach, Coordination, Information Disemmination

Publications

Warren and Les are working on a paper to be submitted to 'Network Support for Grid Computing' (Special Issue of Computer Communications).

Les and Andy are working on a paper for the Edinburgh Global Grid Forum in July on file transfer and compression. Les started making some measurements and wrote code to facilitate.

Les reviewed about 30 papers for Passive & Active Measurement conference and submitted the reviews, since he is on the review committee.

BaBar/HENP

Les prepared and made a presentation for BaBar on measuring high performance (see High Throughput Performance Measurements.

Les worked with IN2P3 and ESnet to understand poor performance to CERN/IN2P3, see http://www.slac.stanford.edu/grp/scs/net/case/cern/problem-20011213.html

Les & Warren met with Sylvan Ravot of Caltech discussed TCP dynamics in particular the slow start threshold, high throughput measurements and QBSS.

Les atttended the ICFA/SCIC meeting at CERN, and prepared and made a presentation (see Network performance measurements). Les also wrote a trip report for the ICFA/SCIC meeting at CERN, see http://www.slac.stanford.edu/grp/scs/trip/notes-icfa-dec01-cottrell.html

Les met with Dominique Boutigny to discuss IN2P3 connectivity performance and ssh issues.

INFN

SLAC hosted a student (Fabrizio Coccetti) from INFN/Trieste. He ported PingER to Trieste and Milan, and developed some infrastructure managment tools to remotely configure IEPM-BW hosts (set up directories, install softwware, set up links), and to build a read and build a database of the remote configurations for all hosts. He also helped write some programs to build scatter plots of iperf vs bbcp disk to disk vs bbcp mem to mem vs bbftp vs pipechar throughputs/bandwidths (see Correlation Plots). he returned to Italy on Xmas Eve.
Back to Top

Crated December 28, 2001.
Comments to iepm-l@slac.stanford.edu