The programming for this lab is done in Java and requires the use of Java datagrams.

Transcription

1 Lab 2 Traffic Regulatin This lab must be cmpleted individually Purpse f this lab: In this lab yu will build (prgram) a netwrk element fr traffic regulatin, called a leaky bucket, that runs ver a real netwrk. The traffic fr testing the leaky bucket will be the Pissn, VBR vide, and LAN traffic traces frm Lab 1. Sftware Tls: The prgramming fr this lab is dne in Java and requires the use f Java datagrams. What t turn in: Turn in a hard cpy f all yur answers t the questins in this lab, including the plts, hard cpies f all yur Java cde, and the annymus feedback frm. Versin 1 (January 27, 2007) Jörg Liebeherr, All rights reserved. Permissin t use all r prtins f this material fr educatinal purpses is granted, as lng as use f this material is acknwledged in all derivative wrks.

2 Table f Cntent Table f Cntent 2 Preparing fr Lab 2 2 Lab 2 Errr! Bkmark nt defined. Part 1. Prgramming with Datagram Sckets and with Files 3 Part 2. Traffic generatrs 5 Part 3. Leaky Bucket Traffic Regulatr 7 Part 4. Regulating a VBR vide surce with Multiple Leaky buckets 10 Feedback Frm fr Lab 2 12 Preparing fr Lab 2 This lab requires netwrk prgramming with Java datagram sckets. There is an infrmative and shrt tutrial n Java datagrams is available at: Java datagram sckets use the UDP transprt prtcl t transmit traffic. The relatinship between Java datagrams and the UDP prtcl is described in: Cmments Quality f plts in lab reprt: This lab asks yu t prduce plts fr a lab reprt. It is imprtant that the graphs are f high quality. All plts must be prperly labeled. This includes that the units n the axes f all graphs are included, and that each plt has a header line that describes the cntent f the graph. Feedback: T be able t imprve the labs fr future years, we cllect data n the current lab experience. Yu must submit an annymus feedback frm fr each lab. Please use the feedback frm at the end f the lab, and return the frm with yur lab reprt. Extra credit: Cmplete the ptinal Part 4 fr 10% extra credit. Java: In the Unix lab, the default versin f the Java installatin is relatively ld. T access a mre recent versin use the cmmand: Cmpiling: /lcal/java/jdk1.5.0_09/bin/javac Running: /lcal/java/jdk1.5.0_09/bin/java ECE 1545 LAB 2 - PAGE 2 J. Liebeherr

3 Part 1. Prgramming with Datagram Sckets and with Files The purpse f this part f the lab is t becme familiar with prgramming Datagram sckets and with writing Java prgrams that read and write data t/frm a file. The prgrams prvided in this part intend t ffer guidance fr the prgramming tasks needed later n. Exercise 1.1 Prgramming with datagram sckets Cmpile and run the fllwing tw prgrams. The prgram Sender.java transmits a string t the receiver ver a datagram scket. The prgram Receiver.java displays the string when it is received. Sender.java Receiver.java imprt java.i.*; imprt java.net.*; public class Sender { public static vid main(string[] args) thrws IOExceptin { InetAddress addr = InetAddress.getByName(args[0]); byte[] buf = args[1].getbytes(); DatagramPacket packet = new DatagramPacket(buf, buf.length, addr, 4444); DatagramScket scket = new DatagramScket(); scket.send(packet); } } imprt java.i.*; imprt java.net.*; public class Receiver { public static vid main(string[] args) thrws IOExceptin { DatagramScket scket = new DatagramScket(4444); byte[] buf = new byte[256]; DatagramPacket packet = new DatagramPacket(buf, buf.length); System.ut.println("Waiting..."); scket.receive(packet); String s = new String(p.getData(), 0, p.getlength()); System.ut.println(p.getAddress().getHstName() + ": " + s); } } Cmpile the prgrams. Start the receiver by running java Receiver. Assuming that the receiver is running n a hst with IP address , start the sender by running: java Sender My String The receiver prgram shuld nw display the string My String. Repeat this exercise, with the difference, that yu run the sender and receiver n tw different hsts. ECE 1545 LAB 2 - PAGE 3 J. Liebeherr

4 Exercise 1.2 Reading and Writing data frm a file Dwnlad the Java prgram ReadFileWriteFile.java. The prgram reads an input file "data.txt" which has entries f the frm I P B The file is read line-by-line, the values in a line are parsed and assigned t variables. Then the values are displayed, and written t a file with name utput.txt. Run the prgram with the VBR vide trace frm Lab 1 as input. The vide trace is available at: Mdify the prgram s that it cmputes and displays the average size f the fllwing frame types: I frames; P frames; B frames. ECE 1545 LAB 2 - PAGE 4 J. Liebeherr

5 Part 2. Traffic generatrs The gal in this part f the lab is t build a traffic generatr that emulates realistic traffic surces. The traffic generatr is driven by a trace file, i.e., ne f the trace files frm Lab 1. The entries in the file permit t determine the size f transmitted packets and the elapsed time between packet transmissins. Fr each entry in the trace file, the traffic generatr creates a UDP datagram f the indicated size and transmits the datagram t the traffic sink. The traffic sink recrds time and size f each incming datagram, and writes the infrmatin int a file. The scenari is depicted in the figure belw. Trace file Time (in ms) size (in bytes) Arrival (in ms) size (in bytes) bytes 300 bytes 13 ms Traffic Generatr Traffic Sink In this part f the lab, yu will build a traffic generatr fr the trace files with Pissn traffic. In Part 3, yu wrk with the Bellcre trace file. In Part 4, yu use the VBR vide traffic. Exercise 2.1 Traffic Generatr fr Pissn traffic Write a prgram that is a traffic generatr fr the cmpund Pissn traffic surce (see Lab 1, Exercise 1.4) and that transmits the traffic using Java datagrams t a traffic sink. The traffic sink has t be prgrammed as well. A trace fr the cmpund Pissn surce is available at URL The file has the frmat: SeqN Time (in μsec) Size (in Bytes) Re-scale the data in the file as fllws: Multiply the time values in the file by a factr f 10 Multiply the packet size by a factr f 10. ECE 1545 LAB 2 - PAGE 5 J. Liebeherr

6 This results in a cmpund Pissn prcess with packet arrival rate λ = 125 packets/sec, and the packet size has an expnential distributin with average size 1/μ =1000 Bytes. The traffic generatr reads each line frm the trace file, re-scales the values, and transmits a UDP datagram packet using the fllwing cnsideratins: The size f the transmitted datagram is equal t the (re-scaled) packet size value; The time f transmissin is determined frm the (re-scaled) time values. (The first packet shuld be transmitted withut delay). Exercise 2.2 Build the Traffic Sink Write a prgram that serves as traffic sink fr the traffic generatr frm the previus exercise. The requirements fr the traffic sink are as fllws: Read packets frm a specified UDP prt; Fr each incming packet, write a line t an utput file that recrds the size f the packet and the time since the arrival f the previus packet (Fr the first packet, the time is zer). Test the traffic sink with the traffic generatr frm Exercise 2.1. Exercise 2.3 Evaluatin Run experiments where yu transmit traffic frm the traffic generatr t the traffic sink. Evaluate the accuracy f the traffic generatr by cmparing the entries in the trace file (at the traffic generatr) t the results written t the utput file (at the sink). Use at least 10,000 data pints fr yur evaluatin. Prepare a plt that shws the difference f trace file and the utput file. Fr example, yu may create tw functins that shw the cumulative arrivals f the trace file and the utput file, respectively, and plt them as a functin f time. Try t imprve the accuracy f the traffic generatr. Evaluate and graph yur imprvements by cmparing them t the initial plt. Exercise 2.4 (Optinal) Accunt fr packet lsses. Packet lsses may ccur due t bit errrs, buffer verflws, cllisins f transmissins, r ther reasns. Packet lsses are less likely if bth the sender and the receiver are n the same machine. The UDP prtcl des nt recver packet lsses. Indicate in yur plts frm the evaluatin any packet lsses. Lab Reprt: Prvide discussins and graphs as requested in Exercise 2.3 and (the ptinal) Exercise 2.4. ECE 1545 LAB 2 - PAGE 6 J. Liebeherr

7 Part 3. Leaky Bucket Traffic Regulatr Yu will design and implement a leaky bucket regulatr with size L and rate r, as shwn in the figure. Tkens are fed int the bucket at rate r. N mre tkens are added if the bucket cntains L tkens. Data can be transmitted nly if there are sufficient tkens in the bucket. T transmit a packet f N bytes, the bucket must cntain at least N tkens. If there are nt sufficient tkens, the packet must wait until there are enugh tkens in the buffer. Initially, the leaky bucket is full and the buffer is empty, i.e., LB(t) = L bytes and X(t) = 0 bytes. Nte that L shuld nt be smaller than the maximum packet size. r L LB(t) A(t) D(t) X(t) Exercise 3.1 Build a Leaky Bucket traffic regulatr Yur task is t build a leaky bucket regulatr that receives data frm a traffic generatr and transmits the utput t a traffic sink. This is illustrated belw. Trace file Time (in ms) size (in bytes) Arrival size X(t) L(t) (in ms) (in bytes) Arrival (in ms) size (in bytes) LB (r, L) Traffic Generatr Leaky Bucket Traffic Sink The transmissins between traffic generatr, leaky bucket, and traffic sink use UDP datagrams. The size and timing f packet transmissins is dne as described in Part 2. Upn each packet arrival, the leaky bucket regulatr writes a line t an utput file that recrds the size f the packet and the time since the arrival f the last packet (Fr the ECE 1545 LAB 2 - PAGE 7 J. Liebeherr

8 first packet, the time is zer). Als recrded are the number f tkens (L(t)) in the leaky bucket and the backlg in the buffer (X(t)) after the arrival f the packet. Exercise 3.2 Regulating LAN traffic frm the Bellcre trace Evaluate the accuracy and crrectness f yur leaky bucket implementatin with the aggregate LAN traffic frm Bellcre as traffic surce. Take at least 10,000 data pints f the Bellcre traffic trace, available at Mdify the entries in the trace file as fllws (This can be dne when the data is read): All packets with a size f 1518 bytes are cunted as having a size f 1480 size (What is the reasn fr this adjustment?). The time units in the first clumn are interpreted as secnds (as ppsed t millisecnds). Determine the average traffic rate r* f yur trace. Using the setup frm Exercise 3.1, set the parameters f the leaky bucket (L, r), such that L = 1480 bytes and r = r*. Transmit packets frm the traffic generatr using the (truncated and mdified) data in the Bellcre trace. Prepare a plt that shws the differences between the trace file and the utput file, as a functin f time. Prepare plts, that depict the size f the leaky bucket L(t) and the backlg in the buffer X(t) as a functin f time. What percentage f time is there a backlg at the leaky bucket? What is the lngest backlg at the leaky bucket? Exercise 3.3 Perfrm a stress test Perfrm a stress test f yur leaky bucket implementatin that determines the maximum transmissin rate that can be supprted by yur leaky bucket Fr the stress test yu will create a cnstant-rate traffic surce that generates fixed-sized packets at cnstant time intervals. When the time between packets is decreased, the rate f data transmissin is increased. Eventually, the leaky bucket will nt be able t keep up and a grwing backlg will be bserved at the buffer f the leaky bucket. The highest rate at which n (permanent and grwing) backlg is bserved marks the maximum transmissin rate. Create a traffic generatr that transmits packets f a fixed size (L bytes) with cnstant inter-packet spacing f T ms. Set up a leaky bucket traffic regulatr with parameters (L, r). Fr a packet size f L = 100 bytes, determine the smallest inter-packet spacing T min that can be supprted by yur implementatin. Start with an initial spacing (chsen ECE 1545 LAB 2 - PAGE 8 J. Liebeherr

9 by yu) and reduce the time between packets, until yu see a lng backlg develping in the buffer f the leaky bucket. Determine the maximum transmissin rate fr scenaris, where bth the sender and the regulatr are running n the same system, and where sender and regulatr are running n tw different cmputers. Repeat the experiment with different packet sizes, and evaluate the impact f the packet size n the maximum transmissin rate. Exercise 3.4 Imprving the implementatin Try t imprve the perfrmance f the leaky bucket traffic generatr, s that it can supprt higher data rates. The fllwing are factrs that may imprve the perfrmance: Avid flating pint peratins. Perfrm peratins in Integer arithmetic, aviding divisins. Avid unnecessary memry allcatins (creatin f bjects) and system calls (e.g., timers). Nte: By implementing the slutin in Java, sme aspects with an impact n the perfrmance are nt under yur cntrl. An imprtant limitatin is that garbage cllectin, i.e., the de-allcatin unused memry, is nt cntrlled by the user. Yur grade in this sectin depends n the maximum rate that yu achieve (in cmparisn t ther implementatins). Yu may be asked t demnstrate the perfrmance in the next lab sessin. Lab Reprt: Describe the design f yur leaky bucket implementatin. Fr Exercise 3.2, prvide a set f plts and include a descriptin f the plts. Fr Exercise 3.3, prvide a plt t supprt yur the data rate the plts, a descriptin f the plts, and yur bservatins and discussins. Fr Exercise 3.4 describe the imprvements t yur cde and the results that yu achieved. Cmpare the results with thse f Exercise 3.3. ECE 1545 LAB 2 - PAGE 9 J. Liebeherr

10 Part 4. (Optinal) Regulating VBR vide with Multiple Leaky buckets The regulatin f VBR vide traffic with a leaky bucket has t address the fllwing tw prblems: Since the size f a frame can be much larger than the maximum packet size (in the file mvietrace.data, the maximum frame size exceeds 600 KB), yet the time lag between frames is relatively lng, ne wuld like t select leaky bucket parameters that stretch the transmissin f a frame ver the entire 33 ms. In rder reduce the lng-term resurce cnsumptin f the VBR vide surce, ne wuld like t set the leaky bucket rate parameter t a value that is clse t the average rate f the VBR surce. T satisfy bth cnstraints, ne leaky bucket is clearly nt sufficient. A pssible (and realized) slutin t this prblem is t regulate VBR vide traffic by tw leaky buckets put in series: The first leaky bucket cntrls the peak rate, by spacing ut the transmissin f a frame ver a lnger duratin; The secnd leaky bucket cntrls the average rate f the VBR surce. An illustratin f a dual leaky bucket is shwn in the figure belw. An arriving packet must withdraw tkens frm bth leaky buckets. If ne f the tw buckets des nt have enugh tken, the packet must wait until sufficient tkens are available in bth buckets. r 1 r 2 L 1 LB1(t) L 2 LB2(t) A(t) D(t) X(t) ECE 1545 LAB 2 - PAGE 10 J. Liebeherr

11 Exercise 4.1 Dual Leaky Bucket Traffic regulatr Implement a traffic regulatr that realizes a dual leaky bucket. The leaky bucket has fur parameters (L 1, r 1, L 2, r 2 ), fr the size and rate f the leaky buckets. As in Exercise 3.1, the regulatr receives data frm a traffic generatr and transmits the utput t a traffic sink. Build a traffic generatr (f yur chice) that can be used t demnstrate the crrectness and accuracy f yur implementatin. Prvide plts (similar t Exercise 3.2) that illustrate hw yu tested the implementatin. Exercise 4.2 Selecting dual leaky bucket parameters fr a VBR vide surce Yur task is t cntrl the VBR vide trace frm Exercise 1.2 with yur dual-leaky bucket implementatin. Transmitted packets are nt permitted t exceed 1480 Bytes. S, sme frames must be divided up int multiple packets. Yur task is t make a `gd selectin fr the parameters f a dual leaky bucket fr the parameters. The selectin f parameters requires yu t tradeff multiple cnsideratins: 1. The maximum buffer size (and the maximum waiting time) shuld nt be t large; 2. The average rate allcatin shuld nt be much larger than the average rate f the VBR surce (t avid ver-allcatin f bandwidth); 3. The number f packets that can be transmitted at nce (this is determined by min(l 1, L 2 )), shuld be small. Transmit the vide surce frm the traffic generatr using the data in file mvietrace.data. Prepare a plt that shws the differences between the trace file and the utput file, as a functin f time. Prepare plts that depict the size f the leaky bucket L(t) and the backlg in the buffer X(t) as a functin f time. What percentage f time is there a backlg at the leaky buckets? What is the lngest backlg at the leaky buckets? Prepare a plt that shws the differences between the trace file and the utput file, as a functin f time. Als prvide a plt that shws the backlg in the buffer X(t) as a functin f time. Determine the lngest backlg X(t) and the maximum waiting time. ECE 1545 LAB 2 - PAGE 11 J. Liebeherr

12 Feedback Frm fr Lab 2 Cmplete this feedback frm at the cmpletin f the lab exercises and submit the frm when submitting yur lab reprt. The feedback is annymus. D nt put yur name n this frm and keep it separate frm yur lab reprt. Fr each exercise, please recrd the fllwing: Part 1. Prgramming with Datagram Sckets and with Files Part 2. Traffic generatrs Difficulty (-2,-1,0,1,2) -2 = t easy 0 = just fine 2 = t hard Interest Level (-2,-1,0,1,2) -2 = lw interest 0 = just fine 2 = high interest Time t cmplete (minutes) Part 3. Leaky Bucket Traffic Regulatr Part 4. Regulating a VBR vide surce with Multiple Leaky buckets Please answer the fllwing questins: What did yu like abut this lab? What did yu dislike abut this lab? Make a suggestin t imprve the lab. ECE 1545 LAB 2 - PAGE 12 J. Liebeherr