These tests have been desgined to determine how fair various transport protocols are. Currently we
are experimenting with:
High-Speed TCP
Hamilton TCP
Scalable TCP
Layered TCP
Reno TCP
HSTCP-LP
UDP based Data Transport (UDT)
Design
We run a volley of 4 TCP/UDP flows, one after the other each separated by an interval of 2 minutes.
Each flow leaves in LIFO (last in first out) order. As shown in the figure below, the test is
broken up into 7 regions, and statistics are collected for each of the seven regions as well as per
individual flows. Aggregate throughput values are also collected for each of the regions as well as
for the overal test.
Flow 1 is represented by RED
Flow 2 by GREEN
Flow 3 by BLUE
Flow 4 by PINK
Parameters
Caltech
UFL
CERN
RTT ms
13.6
78.9
164
Bottleneck Mbps
600-1000
400-600
400-700
Window Size
16MB
16MB
16MB
Also the following line was added in /etc/sysctl.conf for the sender and receiver. These values are
used as default unless specified otherwise:
net.core.rmem_max = 33554432
net.core.wmem_max = 33554432
net.core.rmem_default = 65536
net.core.wmem_default = 65536
net.ipv4.tcp_rmem = 4096 87380 33554432
net.ipv4.tcp_wmem = 4096 65536 33554432
net.ipv4.tcp_mem = 33554432 33554432 33554432
net.ipv4.route.flush = 1