1. Data Transport Challenges for e-VLBI Julianne S.O. Sansa* * With Arpad Szomoru, Thijs van der Hulst & Mike Garret

2. e-VLBI meeting 12 October 20052 Outline Network performance tests Simulation results conclusion

3. e-VLBI meeting 12 October 20053 Network Performance Measurements Investigate critically several connections established. Wire speeds suggests much higher throughput than what application data realises. TCP Congestion Control algorithm (AIMD) –SSACK:Cwnd  Cwnd +1 –CAACK:Cwnd  Cwnd + 1/Cwnd DROP: Cwnd  Cwnd -1/2*Cwnd Cwnd = max. # packets that TCP injects into network before receiving ACK. Cwnd optimal ~ Throughput *RTT Cwnd average = 1.22*MSS/sqrt (p) [Floyd & Fall (1999), Padhya et.al (1998)]

4. e-VLBI meeting 12 October 20054 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)?

5. e-VLBI meeting 12 October 20055 Results with web100 File transfer of 10 GB & 1GB file Modified file transfer (app socket buffers) Memory-memory with bwctl

6. e-VLBI meeting 12 October 20056 Cwnd, RwinRcvd & for a file transfer / memory-memory

7. e-VLBI meeting 12 October 20057 Achieved/Available throughput

8. e-VLBI meeting 12 October 20058 Summary Test results Memory –MemoryFile transferModified file transfer Disk2net-net2file (yet to be done) Cwnd (bytes) TCP (Mbps) UDP (Mbps) Cwnd (bytes) TCP (Mbps) Cwnd (bytes) TCP (Mbps) CwndTCPUDP Bench via Amste rdam 2251640624.19556516056.6220000191.1

9. e-VLBI meeting 12 October 20059 NIC RTT/loss discrepancies

10. e-VLBI meeting 12 October 200510 TCP /Application throughput

11. e-VLBI meeting 12 October 200511 The bottlenecks Application socket buffers Hardware (PCI bus limit, NICs) The OS (more or less tuned optimally) The transport protocol (TCP) –Window limits –Retransmissions

12. e-VLBI meeting 12 October 200512 Transport Protocol Analysis Already many proposals to alter this behaviour: HighSpeed TCP, scalable TCP, Westwood TCP, HTCP, Vegas, FAST, BIC, C-TCP

13. e-VLBI meeting 12 October 200513 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.

14. e-VLBI meeting 12 October 200514 Preliminary Simulation results Simulated file transfer of bench via Amsterdam scenario TCP UDPHSTCPFAST Cwnd (bytes) 1,220.86 n/a4,543.42205 T/put (bps) 61,600,000960,000,00061,600,000310,870,560

15. e-VLBI meeting 12 October 200515 Cwnd for the simulated protocols

16. e-VLBI meeting 12 October 200516 Achieved Throughput for the simulated protocols

17. e-VLBI meeting 12 October 200517 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)

18. e-VLBI meeting 12 October 200518 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)