1. 1 Portable Networks: Prototype and Performance Carey Williamson Guangwei Bai and Kehinde Oladosu Dept of Computer Science University of Calgary

2. 2 Introduction Wireless technologies are prevalent today; continued growth in popularity Example: IEEE 802.11b WLAN (“WiFi”) Economical, convenient, flexible solution for tetherless network access (11 Mbps) Enabler for mobile computing Two possible modes of usage: Infrastructure mode Ad hoc mode 2/36

3. 3 Example: Infrastructure Mode Carey Internet Access Point (AP) cnn.com

4. 4 Example: Ad Hoc Mode Multi-hop “ad hoc” networking Carey Mike

5. 5 Example: Ad Hoc Mode Multi-hop “ad hoc” networking Carey Mike

6. 6 Example: Ad Hoc Mode Multi-hop “ad hoc” networking Carey Mike

7. 7 Example: Ad Hoc Mode Multi-hop “ad hoc” networking Carey Mike

8. 8 Motivation Observation: The same wireless technology that allows clients to be mobile also allows servers to be mobile Hybrid networking paradigm, combining client-server and ad hoc networking, without general Internet infrastructure Portable, short-lived, ad hoc networks “Portable networks” Is this useful? How well does it work? 8/36

9. 9 Portable Network (1 of 2) Carey Access Point (AP) mystuff.com

10. 10 Portable Network (2 of 2) Carey mystuff.com

11. 11 Portable Networks Concept Set up when needed, tear down after Typically needed for minutes or hours When and where not known a priori No existing network infrastructure General Internet access not available, but not required either Pre-defined content; target audience Modest number of users; mobile too

12. 12 Example Usage Scenarios #1 Classroom area network (e.g. “legacy classroom”) Press conferences, media events Conventions and trade shows Disaster recovery sites Recruiting events Schools Voting...

13. 13 #2 Wireless Media Servers

14. 14 #3 Wireless Gaming Multi-player gaming over an ad hoc wireless network

15. 15 #4 Home Networking Even for homes without Internet access, wireless media servers and ad hoc networks could be quite useful Possible usage scenarios: use PDA to read recipe while in the kitchen page your kids for supper time work while on your back porch (backups) music in any room of the house portable media player for parties family gaming fun

16. 16 Research Objectives Assess feasibility of portable networks Benchmark the capabilities and limitations of wireless Web servers in an IEEE 802.11b Wireless LAN Identify performance bottlenecks Understand sensitivity of performance to different workload assumptions Understand impacts of wireless network channel quality and error rates 16/36

17. 17 Experimental Setup wireless sniffer clientsserver... SnifferPro 4.6 Apache 1.3.23 httperf  Empirical measurement methodology Network: In-building 11 Mbps IEEE 802.11b wireless LAN, operating in ad hoc mode (single hop; no mobility) Hardware: Compaq Evo N600c notebooks (1.2 GHz Pentium III, 128 MB RAM, 512 KB L2 Cache, Cisco Aironet 350 network cards Software: Redhat Linux 7.3 (kernel 2.4.18-3), Apache, SnifferPro, httperf for Web workload generation

18. 18 Tutorial: HTTP and TCP TCP is a connection-oriented protocol SYN SYN/ACK ACK GET URL YOUR DATA HERE FIN FIN/ACK ACK Web ClientWeb Server

19. Example Web Page Harry Potter Movies As you all know, the new HP book will be out in June and then there will be a new movie shortly after that… “Harry Potter and the Bathtub Ring” page.html hpface.jpg castle.gif

20. Client Server The “classic” approach in HTTP/1.0 is to use one HTTP request per TCP connection, serially. TCP SYN TCP FIN page.html Get TCP SYN TCP FIN hpface.jpg Get TCP SYN TCP FIN castle.gif Get

21. Client Server Concurrent (parallel) TCP connections can be used to make things faster. TCP SYN TCP FIN page.html Get castle.gif Get F S hpface.jpg S F C S C S

22. Client Server The “persistent HTTP” approach can re-use the same TCP connection for Multiple HTTP transfers, one after another, serially. Amortizes TCP overhead, but maintains TCP state longer at server. TCP FIN Timeout TCP SYN page.html Get hpface.jpg Get castle.gif Get

23. 23 Benchmarking Experiments Factor Levels Number of Clients 1, 2, 4 HTTP Transaction Rate (per-client)10, 20, 30, …, 160 HTTP Transfer Size (KB)1, 2, 4, 8, …, 100 Persistent Connectionsno, yes HTTP Requests per Connection1, 5, 10, 15, …, 60 Transmit Power (mW)1, 5, 20, 30, 50, 100 Client-Server Distance (m)1, 10, 100 Experimental Factors and Levels Performance Metrics: HTTP response time, network throughput 23/36

24. 24 What is the range of sustainable load? Design: Number of Client: 1 HTTP request rate (req/sec): 10, 20, …, 160 HTTP transfer size: 1 KB Non-persistent HTTP Client-server distance: 1 meter (same desk) Research Question

25. 25 Request Rate Results Transmit Queue Length Transmitted Packet ID (b) Client Transmit Queue Transaction Rate (req/sec) Request Rate (req/sec) (a) Successful HTTP Transaction Rate - Maximum sustainable 1 KB HTTP transaction rate for 1 client is about 85 reqs/sec (throughput about 0.9 Mbps) - Beyond this request rate, the client link-layer transmit queue builds up and overflows, losing packets even before they get onto the wireless LAN! - Wireless LAN is bottleneck (802.11b channel access protocol)

26. CSMA-CA + Acknowledgement Carrier Sense Multiple Access with Collision Avoidance * SIFS - Short Inter-Frame Space (approx 28 µs) Every frame is ack’ed - except broadcast and multicast! “Air” is free for DIFS time period (128 usec) Receive ACK back that frame was received intact! send frame source destination others DIFS SIFS All other devices must defer while “air” is busy data ack NAV: defer access

27. IEEE 802.11b Frame Format Long Preamble = 144 bits Interoperable with older 802.11 devices Entire Preamble and 48 bit PLCP Header sent at 1 Mbps 128-bit Preamble (Long) 16 bit Start of Frame Delimiter Signal Speed 1,2,5.5,11 Mbps Length of Payload 16 bit CRC Payload 0-2312 bytes Transmitted at 1 MbpsTransmitted at X Mbps DLL HDR (IP TCP HTTP Data)

28. 28 Research Question Does persistent HTTP help? Design: Number of Clients: 1, 2 HTTP request rate: 10 req/sec HTTP transfer size: 1 KB Persistent HTTP Client-server distance: 1 m (same desk)

29. 29 Results for Persistent Connections  Peak throughput: 3.2 Mbps, 3.5x improvement over non- persistent connection case (0.9 Mbps) for 1 KB transfers  Typically 2 TCP packets per HTTP transaction (vs 10) Throughput (Mbps) HTTP Req/Connection YES!

30. 30 Research Question What is maximum throughput achievable? Design Number of Client: 1 HTTP request rate: 10 req/sec HTTP transfer size (KB): 1, 2, 4, 8, … Non-persistent HTTP Client-server distance: 1 m (same desk)

31. 31 Transfer Size Results TCP Connection Duration (sec) Connection ID Frequency (%) Duration (Sec) 48 KB: Mean Duration: 97 ms Throughput: 4.1 Mbps  bottleneck shifts to the server’s link layer transmit queue 8 KB: Mean Duration: 24 ms Throughput: 2.9 Mbps 1 KB: Mean Duration: 9.7 ms Throughput: 0.9 Mbps

32. 32 Classroom Experiments (Feb’03) 01400 32/36

34. 34 Summary of Results Wireless Web servers can work! Wireless LAN is the bottleneck Bottleneck manifests itself differently, depending on the Web workload client side, for small HTTP transfers server side, for large HTTP transfers unfairness amongst clients if TCP SYN losses network thrashing in some scenarios Persistent HTTP helps a lot!

35. 35 Conclusions Portable networks: a novel paradigm for the use of ad hoc networks Reasonable performance with existing off-the-shelf hardware and software Performance bottleneck at the WLAN manifests itself in interesting ways IEEE 802.11a (55 Mbps) may help Need to explore novel scenarios for the use of this networking paradigm

36. 36 The End Thanks for your attention! Credits: Guangwei Bai Kehinde (Kenny) Oladosu More info: carey@cpsc.ucalgary.ca Questions?