Test1: Solutions. Silvia Giordano ICA, EPFL. t1-1

Transcription

1 Test1: Solutions Silvia Giordano ICA, EPFL t1-1

2 EX1: Delay Consider sending two large files File1 and File2 from host A to host B. The size of the files is F1 and F2=3*F1 bits, respectively. The two hosts are connected by two single point-to-point links both of m meters. Suppose that the propagation speed along both links is s meters/sec. Suppose that the links are uncongested (that is, no queuing delays), and that the link rate of link1 is R1 bps, and the link rate of link2 is R2 bps. A can transmit on both links in parallel. Answer the following questions: link 1 A link 2 B t1-2

3 EX1: Delay 1) Suppose there is no segmentation (the transmission unit is the entire file). If A starts at time t=t 0 at what time, at earliest, B can receive the whole File1? The fastest way is to send File 1 on the link with larger rate, max(r1,r2). The File 1 is received, at earliest, at: t = t0 + m/s + F1/(max(R1,R2)) t 0 transmission delay F1/(max(R1,R2)) File1 max(r1,r2)) F1 A propagation delay m/s B t1-3

4 EX1: Delay 2) If A starts at time t=t 0 at what time, at earliest, B can receive the whole File2? t = t0 + m/s + F2/(max(R1,R2)) t 0 transmission delay F2/(max(R1,R2)) File2 max(r1,r2)) F2 A propagation delay m/s B t1-4

5 EX1: Delay 3) If A starts at time t=t 0 at what time, at earliest, B can receive the whole File1 and the whole File2? t = t0 + m/s + min{ P1, P2)} P1= max{ F1/min(R1,R2), F2/max(R1,R2) } P2 = (F1+F2)/max(R1,R2) transmission delay P1 = max{f1/min(r1,r2), F2/max(R1,R2) } t 0 File1File2 t 0 t=t0 + m/s + min{ P1, P2)} transmission delay P2=(F1+F2)/(max(R1,R2)) File1+ File2 propagation delay m/s A B A propagation delay m/s B t1-5

6 EX1: Delay 4)Suppose that host A segments the file into segments of S=F1 bits each and adds h bits of header to each segment, forming packets of L = h + S bit. In this question, we suppose R1=R2=R. Which is the best transmission strategy for transmitting both File1 and File2? Sequential (first one file, and, when possible, the other one) or parallel (start to transmit the two files at the same time)? Why? t1-6

7 EX1: Delay As S=F1 and R1=R2=R, the 2 strategies are equivalent: Sequential: start sending File 1. As S=F1, there is only one segment, which is sent on one link. Thus, the other link can be used to start sending File 2. When File 1 ends, we can start using the 2 links in parallel: t = 2L/R + t0 + m/s link1 link2 t1-7

8 EX1: Delay As S=F1 and R1=R2=R, the 2 strategies are equivalent: Parallel: start sending File 1 on one link and File 2 on the other link. When File 1 ends, we can start using both links for File 2: t = 2L/R + to + m/s link1 link2 t1-8

9 EX2: JAVA Consider the Intranet INTRA007 in the diagram 3, with addresses /24. Security is a high concern of INTRA007 and people from inside can go out only with SMTP protocol ( ). A INTRA007 INTERNET B G t1-9

10 EX2: JAVA The gateway G is the access node to/from the Internet. When a person from inside wants to send an , he accesses the INTRA007-userinterface that passes the user s data to the INTRA007- client. The INTRA007- Client builds, with the user data, the following packet, and then it sends it. A INTRA007 ToG INTERNET B G t1-10

11 EX2: JAVA 1) Describe what happens to the mail written by the users before eventually getting to the Internet. The mail is controlled by G. It checks both the recipients and the data of the mail. 2) Describe what is G. G is a controller working on UDP. Technically, it is called an application gateway (will see it next week). t1-11

12 EX2: JAVA 3) Suppose A, B and C clients contact G at the same time and they enter as following in the incoming request queue on the host that runs G: [A, C, B]. At time t, when G receives the first packet, what is going on at A, B and C? A, B and C have sent their packet to G. They have closed the UDP socket and they will not wait for any ACK. The packet from A is being received. After the execution of the datagrampacket.receive for packet sent by A, the packet the packet sent by C is received, and finally, so is the one sent by B. t1-12

13 EX2: JAVA 4) How would you introduce the access to the Web (HTTP), respecting the same security-based philosophy? (You are not required to write any code, just to describe the architecture and how it works). When a user wants to access a web page from/at INTRA007, he accesses the INTRA007.userInterface that passes the user s data to INTRA007.web. This controls the data and check whether the client/server are authorised. If so, the connection is started. t1-13

14 EX3: GBN and SR Consider the Go-back-N and Selective Repeat protocols. Suppose the sequence number space is, for both, of size k. Assume that the round-triptime between the sender S and the receiver R is constant and equal to RTT seconds. sender S pkt0 pkt1 receiver R constant RTT t1-14

15 EX3: GBN and SR 1) What is the largest allowable sender window for Go-back-N that will avoid the problem of understanding if a packet is a new packet or a retransmission, as we analyzed in class? The sender window has to be smaller than k/2. 2) What is the largest allowable sender window for Selective Repeat that will avoid the same problem? k/2 t1-15

16 EX3: GBN and SR 3)Suppose that the whole session consists of h packets and that the sender S starts to send the first packet at time t=t 0. Suppose that the packet pkt2 is lost, and that there are no other losses afterwards. In this question, we suppose that the transmission delay of a packet is α seconds and that the retransmission timer (timeout) starts when sender S starts transmitting the packet. At what time, at earliest, the receiver R can receive the last packet pkth if Go-Back-N protocol is used? t1-16

17 EX3: GBN and SR The pkt0 and pkt1 are correctly transmitted, with pkt2, the timer is started at time t=2α. When it expires, all packets from pkt2 must be retransmitted. To this time we must add the time for propagating a packet to B (rtt/2): t = t0 + rtt/2 + 2α + timeout + (h-2)α = t0 + rrt/2 + timeout + hα S pkt0 pkt1 R timeout pkt2 pkt2 pkt3 t1-17

18 EX3: GBN and SR 4) Suppose that the whole session consists of h packets and that the sender S starts to send the first packet at time t=t 0. Suppose that the packet pkt2 is lost, and that there are no other losses afterwards. At what time, at earliest, the receiver R can receive the last packet pkth if Selective Repeat protocol is used? t1-18

19 EX3: GBN and SR All packets, but pkt2, are correctly transmitted. When timer expires, only pkt2 is retransmitted. To this time we must add the time for propagating a packet to B (rtt/2): t = t0 + rtt/2 + 2α + max(timeout,(h-2) α) + α S pkt0 pkt1 R S pkt0 pkt1 R timeout pkt2 pkt2 pkt3 pkt4 pkt5 timeout pkt2 pkth pkt3 pkt4 pkt5 pktk pkt2 pkt(k+1) t1-19