Leonard Kleinrock Professor, UCLA Computer Science Dept Founder & Chairman, Nomadix Inc

Transcription

1 Leonard Kleinrock Professor, UCLA Computer Science Dept Founder & Chairman, Nomadix Inc SIGCOMM Tutorial August 31, 1999

2

3 July 3, 1969 UCLA will become the first station in a nationwide computer network

4 ª

5 As of now, computer networks are still in their infancy, says Dr. Kleinrock, But as they grow up and become more sophisticated, we will probably see the spread of computer utilities, which, like present electric and telephone utilities, will service individual homes and offices across the country. ª

6 What It Looked Like in 1969 Before September 1969

7 What It Looked Like in 1969 The Interface Message Processor IMP September 1969

8 The IMP 1969

9 The IMP 1969

10 Today

11 What It Looked Like in 1969 October 1969 October 1969

12

13 Who Had the Foresight to Keep This Log? Jon Postel

14

15 First Message on the Internet - ever!

16 But What WAS the First Message Ever Sent on the Internet? Was it What hath God Wrought? Was it This is a Giant Step for Mankind? It was simply a LOGIN from the UCLA computer to the SRI computer. We sent an L - did you get the L? YEP! We sent an O - did you get the O? YEP! We sent a G - did you get the G?

17 But What WAS the First Message Ever Sent on the Internet? SRI Was it What hath God Wrought? Was it This is a Giant Step for Mankind? It was simply a LOGIN from the UCLA computer to the SRI computer. We sent an L - did you get the L? YEP! We sent an O - did you get the O? YEP! We sent a G - did you get the G? UCLA

18 But What WAS the First Message Ever Sent on the Internet? SRI Was it What hath God Wrought? Was it This is a Giant Step for Mankind? It was simply a LOGIN from the UCLA computer to the SRI computer. We sent an L - did you get the L? YEP! We sent an O - did you get the O? YEP! We sent a G - did you get the G? UCLA

19 What It Looked Like in 1969 UCLA serves the Network Measurement Center September November December October The job is to stress the net to its breaking point!

20 Response Time Throughput Loss RESPONSE TIME T INPUT CAPACITY THROUGHPUT 0 ª Leonard LOSS Kleinrock 1999 COMPUTER NET

21 Flow Control Output 0 0 IDEAL DYNAMIC CONSERVATIVE FREE-FLOW 0 Input DEADLOCK

22 FLOW CONTROL PING! BIG! DP SYS COMPUTER NET

23 FLOW CONTROL Whoa!! BIG! DP SYS COMPUTER NET

24 FLOW CONTROL BIG! DP SYS COMPUTER NET

25 Response Time vs Throughput POWER = Do you want to operate Or here? T( ) Or here? Throughput Response Time = P = T( ) Response Time 0 1/P * Throughput Max Power Point ( )

26 Response Time vs Throughput T( ) POWER = P = Throughput Response Time T( ) Response Time 0 1/P Throughput ( )

27 Response Time vs Throughput T( ) At Max Power N * = 1 Response Time Max Power Point 0 * Throughput ( )

28 N*=1 T( ) M/G/1 0

29 Use Your Intuition Only 1 customer Morale: Just keep the pipe full! T = Min Eff = Max

30 A Brief History of Radio Marconi 1890 s

31 A Brief History of Radio Marconi 1890 s

32 A Brief History of Pkt Radio 1970 s: ARPA 250 cu in 25 watts 25 pounds

33 A Brief History of Pkt Radio 1990 s: ARPA 10 cu in 1 watt 1 pound

34

35

36

37 Giant Stepping in Packet Radio Multihop Each hop covers distance R (Tx Radius) Total distance to cover is D (D>>R) Big R, more interference, fewer hops Small R, less interference, more hops Total Delay = T(R)[D/R D/R] Choose R=R * to minimize total delay dt(r)/dr = T(R)/R optimality condition

38 dt(r)/dr = T/R T(R) R * R

39 Hit any key to continue No.. Not that one!!!

40 Thank You