The Utility of Characterizing Packet Loss as a Function of Packet Size in Commercial Routers
0 likes621 views
The document investigates the correlation between packet size and packet loss in commercial routers, highlighting that buffer implementation significantly impacts packet loss rates. It presents a methodology for characterizing packet loss relative to packet size, which can inform decisions related to network efficiency and service performance. The findings indicate that byte-sized buffers increase packet loss with larger sizes, while packet-sized buffers show consistency in loss rates.
The Utility of Characterizing Packet Loss as a Function of Packet Size in Commercial Routers
- 2. The Utility of Characterizing Packet Loss as a Function of Packet Size in Commercial Routers Jose Saldana Julián Fernández-Navajas José Ruiz-Mas Eduardo Viruete Navarro Luis Casadesus Communication Technologies Group (GTC) Aragon Institute of Engineering Research (I3A) University of Zaragoza, Spain
- 3. INDEX I. Introduction and Related Works II. Tests and Results III. Conclusions
- 4. INDEX I. Introduction and Related Works II. Tests and Results III. Conclusions
- 5. CCNC 2012. Las Vegas, 15 Jan 2011 Introduction and Related Works - The Internet was designed as a best-effort network - The first deployed services were e-mail, ftp, etc., which did not have critical delay requirements. - So they used big packets in order to have a good bandwidth efficiency.
- 6. CCNC 2012. Las Vegas, 15 Jan 2011 Introduction and Related Works - Nowadays, real-time services have very critical delay requirements - So they require a high rate of packets, which implies small sizes
- 7. CCNC 2012. Las Vegas, 15 Jan 2011 Introduction and Related Works - Efficiency problem One IPv4/TCP packet 1500 bytes η=1460/1500=97% One IPv4/UDP/RTP packet of VoIP with two samples of 10 bytes η=20/60=33%
- 8. CCNC 2012. Las Vegas, 15 Jan 2011 Introduction and Related Works - Packet loss: one of the main impairments in current networks - Wireless: The radio link is the critical one - Wired: Dropping in router queues - TCP-based services can recover lost packets - UDP-based services: interactive applications can be seriously affected, as there is no time for retransmission
- 9. CCNC 2012. Las Vegas, 15 Jan 2011 Introduction and Related Works - ¿Is there a relationship Packet loss vs. Packet size? - Wireless environment: YES. The bigger the packet, the bigger the probability of a corrupted bit. - Wired environment: The main cause of packet loss is dropping in router queues, so we have to study the behavior of the queues. - Many routers were designed for “classical” services, so they are not prepared to deal with small packets
- 10. CCNC 2012. Las Vegas, 15 Jan 2011 Introduction and Related Works - Buffers of commercial routers have different implementations - Buffers can be measured in bytes, packets, or maximum queuing delay - Bandwidth is not the only limit: packets per second limit is also present - drop-tail, RED, etc - The implementation may not strongly affect delay and jitter, but may affect packet loss
- 11. CCNC 2012. Las Vegas, 15 Jan 2011 Introduction and Related Works - Characterization of packet loss vs packet size packet loss packet size
- 12. CCNC 2012. Las Vegas, 15 Jan 2011 Introduction and Related Works - Useful to make some decisions: - Multiplexing or not - Samples per packet - Number of TS (Transport Stream) included in a packet One TS per packet probability packet size
- 13. CCNC 2012. Las Vegas, 15 Jan 2011 Introduction and Related Works - Useful to make some decisions: - Multiplexing or not - Samples per packet - Number of TS (Transport Stream) included in a packet Five TS per packet probability packet size
- 14. CCNC 2012. Las Vegas, 15 Jan 2011 Introduction and Related Works What is better? More TS per packet increases efficiency But if it increases packet loss… Five TS per packet probability packet size
- 15. CCNC 2012. Las Vegas, 15 Jan 2011 Introduction and Related Works - So - Let’s study packet loss vs. packet size behavior of the router (and the Internet, as future work) - Help when making decisions - Give us an idea of the internal (and nonpublicly available) router implementation - The interest is the shape of the graphs
- 16. CCNC 2012. Las Vegas, 15 Jan 2011 Introduction and Related Works “Sizogram” packet loss packet size
- 17. INDEX I. Introduction and Related Works II. Tests and Results III. Conclusions
- 18. CCNC 2012. Las Vegas, 15 Jan 2011 Tests and Results - Definition of a test able to reflect this - An “empirical” test, suitable to be performed with different routers and networks - The router will be considered a “black box” - We do not know its internal structure
- 19. CCNC 2012. Las Vegas, 15 Jan 2011 Tests and Results - Background traffic: 100% of the capacity - 50 % packets - 10% packets - 40% packets 40 bytes 576 bytes 1,500 bytes
- 20. CCNC 2012. Las Vegas, 15 Jan 2011 Tests and Results Offered traffic 10000 variable size 1500 bytes 576 bytes 40 bytes 8000 6000 kbps Background 10Mbps 4000 2000 0 100 200 300 400 500 600 700 800 900 1000 1100 1200 1300 1400 1500 Packet size
- 21. CCNC 2012. Las Vegas, 15 Jan 2011 Tests and Results Offered traffic 10000 8000 variable size Variable size 100kbps kbps 6000 4000 2000 0 100 200 300 400 500 600 700 800 900 1000 1100 1200 1300 1400 1500 Packet size 1500 bytes 576 bytes 40 bytes
- 22. CCNC 2012. Las Vegas, 15 Jan 2011 Tests and Results - Test traffic: two orders of magnitude smaller Offered traffic 10000 variable size 1500 bytes 576 bytes 40 bytes 8000 kbps 6000 4000 2000 0 100 200 300 400 500 600 700 800 900 1000 1100 1200 1300 1400 1500 Packet size
- 23. CCNC 2012. Las Vegas, 15 Jan 2011 Tests and Results - Duration of each test: - From 200 to 3,000 seconds for each point, in order to have 25,000 packets of variable size traffic - Matlab simulations
- 24. CCNC 2012. Las Vegas, 15 Jan 2011 Tests and Results - Two buffer implementations: - byte-sized: 10kB, 20kB, 50kB, 100kB - packet-sized: 16, 33, 83, 166 packets - Can be considered “tiny buffers”
- 25. CCNC 2012. Las Vegas, 15 Jan 2011 Tests and Results: Byte-sized packet loss, buffer byte-sized 10% 10kB 20kB 50kB 100kB 9% 8% 7% packet loss 6% 5% 4% 3% 2% 1% 0% 100 200 300 400 500 600 700 800 900 1000 1100 1200 1300 1400 1500 packet size (bytes)
- 26. CCNC 2012. Las Vegas, 15 Jan 2011 Tests and Results: Byte-sized packet loss, buffer byte-sized 10% 10kB 20kB 50kB 100kB 9% Monotonically increasing 8% 7% packet loss 6% 5% 4% 3% 2% 1% 0% 100 200 300 400 500 600 700 800 900 1000 1100 1200 1300 1400 1500 packet size (bytes)
- 27. CCNC 2012. Las Vegas, 15 Jan 2011 Tests and Results: Byte-sized packet loss, buffer byte-sized 10% 10kB 20kB 50kB 100kB 9% Bigger packet – Bigger packet loss 8% 7% packet loss 6% 5% 4% 3% 2% 1% 0% 100 200 300 400 500 600 700 800 900 1000 1100 1200 1300 1400 1500 packet size (bytes)
- 28. CCNC 2012. Las Vegas, 15 Jan 2011 Tests and Results: Byte-sized packet loss, buffer byte-sized 10% 10kB 20kB 50kB 100kB 9% 8% The slope grows as the buffer gets smaller 7% packet loss 6% 5% 4% 3% 2% 1% 0% 100 200 300 400 500 600 700 800 900 1000 1100 1200 1300 1400 1500 packet size (bytes)
- 29. CCNC 2012. Las Vegas, 15 Jan 2011 Tests and Results: Packet-sized packet loss, buffer packet-sized 10% 16packets 33packets 83packets 166packets 9% No variation with packet size, as expected 8% 7% packet loss 6% 5% 4% 3% 2% 1% 0% 100 200 300 400 500 600 700 800 900 1000 1100 1200 1300 1400 1500 packet size (bytes)
- 30. CCNC 2012. Las Vegas, 15 Jan 2011 Tests and Results: pps limit packet loss, buffer byte-sized 50% 10 kb 10kB, 2000pps 45% 10kB, 2500pps 10kB, 3000 pps 40% 10kB, 3500pps 35% packet loss 30% 25% 20% 15% 10% 5% 0% 100 200 300 400 500 600 700 800 900 1000 1100 1200 1300 1400 1500 packet size (bytes)
- 31. CCNC 2012. Las Vegas, 15 Jan 2011 Tests and Results: pps limit packet loss, buffer byte-sized 50% 10 kb 10kB, 2000pps 45% Offered 1,970 pps 10kB, 2500pps 10kB, 3000 pps 40% 10kB, 3500pps 35% packet loss 30% 25% 20% 15% 10% 5% 0% 100 200 300 400 500 600 700 800 900 1000 1100 1200 1300 1400 1500 packet size (bytes)
- 32. CCNC 2012. Las Vegas, 15 Jan 2011 Tests and Results: pps limit packet loss, buffer byte-sized 50% 10 kb 10kB, 2000pps 45% Increases significantly with respect to no pps limitation 10kB, 2500pps 10kB, 3000 pps 40% 10kB, 3500pps 35% packet loss 30% 25% 20% 15% 10% 5% 0% 100 200 300 400 500 600 700 800 900 1000 1100 1200 1300 1400 1500 packet size (bytes)
- 33. CCNC 2012. Las Vegas, 15 Jan 2011 Tests and Results: processing time packet loss, buffer byte-sized 10kB 20% 10kB, tproc=20us 10kB, tproc=40us 18% 10kB, tproc=60us 10kB, tproc=80us 16% 10kB, tproc=100us 14% packet loss 12% 10% 8% 6% 4% 2% 0% 100 200 300 400 500 600 700 800 900 1000 1100 1200 1300 1400 1500 packet size (bytes)
- 34. CCNC 2012. Las Vegas, 15 Jan 2011 Tests and Results: processing time packet loss, buffer byte-sized 10kB 20% 10kB, tproc=20us Graphs have the same shape 10kB, tproc=40us 18% 10kB, tproc=60us 10kB, tproc=80us 16% 10kB, tproc=100us 14% packet loss 12% 10% 8% 6% 4% 2% 0% 100 200 300 400 500 600 700 800 900 1000 1100 1200 1300 1400 1500 packet size (bytes)
- 35. CCNC 2012. Las Vegas, 15 Jan 2011 Tests and Results: processing time packet loss, buffer byte-sized 10kB 20% 10kB, tproc=20us Packet loss increase is not so severe 10kB, tproc=40us 18% 10kB, tproc=60us 10kB, tproc=80us 16% 10kB, tproc=100us 14% packet loss 12% 10% 8% 6% 4% 2% 0% 100 200 300 400 500 600 700 800 900 1000 1100 1200 1300 1400 1500 packet size (bytes)
- 36. INDEX I. Introduction and Related Works II. Tests and Results III. Conclusions
- 37. CCNC 2012. Las Vegas, 15 Jan 2011 Conclusions - We have studied the relationship between packet size and packet loss in routers - Depends on buffer implementation - A methodology has been defined to find the packet loss histogram as a function of packet size - This can give us an idea of the internal implementation of the router, and help us when making some decisions
- 38. CCNC 2012. Las Vegas, 15 Jan 2011 Conclusions - Byte-sized buffers increase loss with size - Packet-sized buffers do not vary - Extension of this to Internet paths and commercial routers