SLAC logo

Througput Performance between SLAC and IN2P3 Network logo

Les Cottrell. Page created: June 12, 2000, last update: June 12, 2000.

Central Computer Access | Computer Networking | Network Group | ICFA-NTF Monitoring 		SLAC Welcome
Highlighted Home
Detailed Home
Search
Phonebook

Introduction

Dominique Boutigny has been making measurements of the FTP throughput between CERN and IN2P3. He observed that the best FTP performance sending 10 streams of data from a bbftp client running on datamove3.slac.stanford.edu (Sun 420 with 4 336MHz cpus and 1GByte of memory and a Gbyte Ethernet interface, running Solaris 2.6) to ccobsn04.in2p3.fr (Sun 450 running Solaris 2.6) was about 7.4Mbps with compression. 
From datamove3 :
 -  2 threads No Comp  7.5 Mb/s
 -  4 threads No Comp 12.4 Mb/s
 -  5 threads No Comp 13.6 Mb/s
 -  6 threads No Comp 19.1 Mb/s
 -  8 threads No Comp 20.1 Mb/s
 - 10 threads No Comp 19.9 Mb/s
 
 -  5 threads with Comp 13.6 Mb/s
 -  8 threads with Comp  6.3 Mb/s
 - 10 threads with Comp  7.4 Mb/s
 
 From tersk02 :
 - 10 threads with Comp 26.4 Mb/s - Second try : 25.1 Mb/s
 
 From shire01 :
 - 10 threads with Comp 37.9 Mb/s
So it appears that the throughput is limited by the machine itself when we use compression. Dominique has seen the load on datamove3 increasing from ~5 up to ~17 when he was trying to use 10 threads with compression. The normal load on datamove3 stays below 1.3, and typically it is much lower. Randy Melen sees CPU usage typically less than 10%. He concludes that nothing is being done on this system whenever he watches it!

Since bulk data transfers from SLAC to other sites are critical to BaBar it is important to understand and resolve this.

Routes

We measured round trip time and loss to the routers along the route using pingroute.pl. We measured the bottleneck bandwidths on 6/12/00 using pathchar:
[ Feedback | Reporting Problems ]
Page owner: Les Cottrell