180 likes | 252 Views
Data Transport Challenges for e-VLBI. Julianne S.O. Sansa*. * With Arpad Szomoru, Thijs van der Hulst & Mike Garret. Outline. Network performance tests Simulation results conclusion. Network Performance Measurements.
E N D
Data Transport Challenges for e-VLBI Julianne S.O. Sansa* * With Arpad Szomoru, Thijs van der Hulst & Mike Garret
Outline • Network performance tests • Simulation results • conclusion e-VLBI meeting 12 October 2005
Network Performance Measurements • Investigate critically several connections established. Wire speeds suggests much higher throughput than what application data realises. • TCP Congestion Control algorithm (AIMD) • SS ACK:Cwnd Cwnd +1 • CA ACK:Cwnd Cwnd + 1/Cwnd • DROP: Cwnd Cwnd -1/2*Cwnd • Cwnd = max. # packets that TCP injects into network before receiving ACK. • Cwndoptimal ~ Throughput *RTT • Cwndaverage = 1.22*MSS/sqrt (p) [Floyd & Fall (1999), Padhya et.al (1998)] e-VLBI meeting 12 October 2005
Specific Questions • How much bandwidth is available to the these TCP connections? Is it what is seen by the app? • If it is less than the theoretic available b/w, what is the bottleneck? • How do we minimise this bottleneck? • What is the stability of these TCP connection (repeatability /predectability)? e-VLBI meeting 12 October 2005
Results with web100 • File transfer of 10 GB & 1GB file • Modified file transfer (app socket buffers) • Memory-memory with bwctl e-VLBI meeting 12 October 2005
Cwnd, RwinRcvd & for a file transfer / memory-memory e-VLBI meeting 12 October 2005
Achieved/Available throughput e-VLBI meeting 12 October 2005
Summary Test results e-VLBI meeting 12 October 2005
NIC RTT/loss discrepancies e-VLBI meeting 12 October 2005
TCP /Application throughput e-VLBI meeting 12 October 2005
The bottlenecks • Application socket buffers • Hardware (PCI bus limit, NICs) • The OS (more or less tuned optimally) • The transport protocol (TCP) • Window limits • Retransmissions e-VLBI meeting 12 October 2005
Transport Protocol Analysis • Already many proposals to alter this behaviour: HighSpeed TCP, scalable TCP, Westwood TCP, HTCP, Vegas, FAST, BIC, C-TCP e-VLBI meeting 12 October 2005
Loss-based, delay-based,or equation-based? • Which way do we go? • Consider getting the best out each world/Allow the application to dynamically detect network conditions & decide which algorithm to use. e-VLBI meeting 12 October 2005
Preliminary Simulation results e-VLBI meeting 12 October 2005
Cwnd for the simulated protocols e-VLBI meeting 12 October 2005
Achieved Throughput for the simulated protocols e-VLBI meeting 12 October 2005
Conclusions & further work • Hardware (PCI bus, NICs,) on end systems as well as the application (buffers) need to be optimised. • Model TCP data flows & relate flow analysis with correlation. • More simulation work on Transport Protocol analysis (response function) e-VLBI meeting 12 October 2005
References • Floyd & Fall (1999) “Promoting the use of end-to-end congestion control in the internet”, IEEE/ ACM Trans. on Networking, August 1999. • Padhya et.al (1998) “Modeling TCP throughput: A Simple model and its empirical validation” in Proc ACM SigCOMM 1998 • Antony et.al(2004) “Exploring Practical Limitations of TCP over Transatlantic Networks” submitted Elsevier Science(2004) e-VLBI meeting 12 October 2005