Name: SSN: Signature: Open book, open notes. Your work must be your own. Assigned seating. Test time: 7:05pm to 8:05pm. You may not use a calculator or PalmPilot to calculate subnetting/host/netid information. You may use a calculator to convert from binary to decimal and back. All questions graded. 1. From Forouzan section 3.3, we know that a switched network consists of a series of interlinked nodes, called switches. These help ensure that only intended recipients receive copies of frames. The text discussed three general approaches: circuit, packet and message switching. Packet switching can be further divided into datagram switching and virtual circuit switching. The following four diagrams show how a layer two network might respond to the need to transmit a single large message from T to R through three intermediate nodes, A, B, and C. Do not assume that A, B, and C are routers. The message to be sent from T to R is the same in all four pictures. Label each of the four diagrams below to illustrate which one of the four switching schemes they illustrate. This is not a question about error-checking, or acknowledgments of packet transmissions assume that the system was designed assuming 100% reliability and that there are no errors on the network. Page 1 of 9
T A B C R T A B C R t a.: Datagram Packet Switching no call-request or call-accept packets to establish a path between T and R, but the message is divided up into multiple packets. (2 points) T A B C R Call Request Call Accept Acknowledgment t b.: Virtual Circuit Packet Switching resembles circuit switching with call setup, but messages are broken into individual packets for travel on shared data network. (2 points) Page 2 of 9
T A Call Request t B C R Call Accept Acknowledgment c.: Circuit Switching establish a path, and then send message in its entirety on a dedicated network. (2 points) T A B C R t d.: Switching message sent hop-by-hop, but there is no breaking up of message into individual packets. (2 points) e. Which of these four approaches most closely models the Internet s implementation of IP? Circle the correct answer. 1. Circuit Switching 2. Datagram Packet Switching 3. Virtual Circuit Packet Switching 4. Switching Answer: 2. If you answered 3, this is correct for TCP, but not for IP. (2 points) Page 3 of 9
2. Find the netid and hostid of the following IP addresses. Assume classful addressing. Remember that the netid is a fragment of an IP address such as 10.5, not a complete IP address like 10.5.0.0. 114.34.2.8 114 34.2.8 (2 points) 172.35.14.9 172.35 14.9 (2 points) 128.24.27.17 128.24 27.17 (2 points) 245.34.21.5 none Special Use none Special Use (2 points) 9.11.67.3 9 11.67.3 (2 points) 3. A host with IP address 128.23.67.3 sends a message to a host with IP address 128.67.23.5. Assume classful addressing. Yes or No: does the message travel through a router? How do you know? Yes these are class B addresses, but the first sixteen bits (first two decimal numbers) are different. Therefore they belong to different networks and must travel through a router. (2 points) 4. What two fundamental properties (not variables) must all routing tables guarantee and what do they mean? Be brief answer in the space provided. 1. Universal routing all valid destination addresses specified in packets arriving at a router must correspond to a route in the routing table. (2 points) 2. Shortest path the next hop for a packet indicated by a route must be the shortest one possible among the paths available to the router (2 points) What two values must a routing or switching table contain? What layer of the protocol stack does each one represent? 1. Destination layer 3 (or in general, layer n) (2 points) 2. Port or Hop layer 2 (or in general, layer n-1) (2 points) Page 4 of 9
5. The ARI organization has received a class C IP address for its internetwork: 192.16.12.0/24. The network manager needs to decide how to assign addresses to each part of the network shown below. This class C network is the only set of addresses available, and the entire set of reachable networks is shown in the diagram below. There are Ethernet LANs in four cities: Jax, San Fran, Tampa, and Dallas. There are 10 users with PCs on the Jax LAN. There are 5 users with PCs on each of the other three LANs. Each LAN also has a router to connect it to the other sites. Sites are interconnected with three point-to-point connections between the Jax routers and each of the other routers. The network manager wishes to suppress the forwarding of broadcasts between networks. Assume every network segment uses IP and routing. 10 users Router Jax T1 T1 T1 San Fran Router Tampa Router Router Dallas 5 users 5 users 5 users Part I: How many total IP addresses will you assign to devices on the Jax LAN? Circle the correct answer: a) 10 b) 11 10 users + 1 router (1 point) c) 14 d) 16 How many total IP addresses will you assign to all devices on the Dallas LAN? a) 5 b) 6 5 users + 1 router (1 point) Page 5 of 9
c) 8 d) 9 e) 10 How many subnets will you assign to this network? a) 0 b) 1 c) 4 d) 5 e) 7 4 subnets for LANs, 3 for WANs (1 point) f) 8 Draw a circle on the diagram for each subnetwork you would create. Draw carefully! (4 points) Based on the answers above, how many bits should we take from the fourth octet of the class C network to describe subnets? Plan for growth of one new LAN and one PC per LAN, but pick the smallest number of subnets that will solve the problem. Assume we may not assign or use the subnet that contains the Class C network number or the Class C broadcast address. a) 0 b) 1 c) 2 d) 3 e) 4 (1 point) If you chose 3 bits and it was consistent with the number of subnets you chose, you received credit. f) 5 Does this subnetting solution make reasonably efficient use of addresses on all network segments for this company? Circle the correct answer and very briefly explain why. No. 16 addresses used to provide subnet for LAN with 10 PCs leaves 14 of 16 unused on WAN links requiring 2. (2 points) Calculate the ratio of the number of addresses that are available for assignment to devices / the total number of addresses possible in the Class C network. Assume that we reserve network and broadcast addresses they are not available for assignment. You do not have to calculate the percentage a formula is sufficient. (2^4 2) subnets * (2^4 2) assignable IPs per subnet / 2^8 possible IPs (for our class C) (2 points) Part II: Suppose now that the number of users on the JAX LAN increases to 58. What subnet mask would you now use on their class C network to allow for a subnet of 58 users? Page 6 of 9
255.255.255.192 (2 points). 58 users requires a subnet of 64 addresses, which in turn requires 6 bits for host ID. This leaves 2 bits for network ID, 11000000 which is.192. Would using this new subnetwork mask produce a sufficient number of assignable subnets for the rest of ARI s network? How many subnets would your mask allow? No. (2 points). Two bits creates 4 combinations, 2 of which cannot be used, leaving 2 assignable across 4 LANs and 3 WANs. Not enough. What subnet mask makes the most sense to use for the T1 (serial) links assuming we cannot assign a subnet with the network number or the broadcast address? 255.255.255.252 (2 points). T1 links are point-to-point, not a bus, and can never have more than 2 devices. Thus there is no need for growth later. 2 devices requires a subnet of 4 combinations, since 2 are wasted, leading to 2 bits for host id. This leaves 6 for network id, 11111100, which is.252. What subnetting scheme did we discuss in class that would allow you to still use this class C network and provide a sufficient number of assignable subnets and host IPs? VLSM or Variable-Length Subnet Masking (2 points). CIDR is what the ISP does to provide you with contiguous blocks of Class Cs. That is not the situation here. 6. Suppose we have a department with n computers on a single Ethernet segment running TCP/IP applications. Traffic between machines is distributed evenly, and broadcasts account for a small percentage of traffic. There are too many collisions and you want to divide the network into two parts. What device would you add to this network to help reduce the number of collisions. Explain how this would help reduce the number of collisions. The one and only answer is Bridge (10 points). It physically divides the LAN into two LANs, forwarding packets that must be sent from one LAN to another, filtering all others. Since the traffic pattern is distributed evenly between devices, this reduces the traffic by 50%, and therefore reduces the number of packets per second on each segment, thus reducing collisions. 7. Why is a repeater not the same thing as an amplifier if they both operate at layer one? An amplifier amplifies the waveform, including any noise that has been introduced in the signal in its journey. A repeater converts the incoming analog waveform back into a digital stream that the analog waveform represented, and re-encodes it as a new, noise-free waveform for delivery. (10 points) Page 7 of 9
8. You are hired as a network assistant and are given the task of configuring a new workstation for use on the network. You are told to assign the workstation the IP address 128.249.27.17/23. Assume that we are using subnetting, and that the network and broadcast addresses may not be assigned to a host. What is the network address (not the netid) for this particular subnetwork? 128.249.26.0 (2 points). /23 means 23 bits of network ID. Thus, the first 23 bits of the 32 in the IP mask are 1s, the rest 0s. The last two octets in binary look like this: 11111110.00000000. Now we apply this mask to 128.249.27.17, the IP address given. The last 16 bits given are 00011011.00010001, which when XOR d with the mast give you 00011010.00000000, or.26.0. What is the range of IP addresses in this network or subnetwork assignable to hosts? 128.249.26.1 128.249.27.254 (2 points). Simply state the range of combinations in the host portion of the network address from all 0 s to all 1 s, removing the first and the last. In other words, from 0.00000001 to 1.11111110. Add this to the network number 128.249.26.0 and you have the range 128.249.26.1 128.249.27.254. It is more difficult to calculate this without converting to binary first. What subnet mask should this workstation use? Show your answer in decimal. 255.255.254.0(2 points). Set all network bits to 1, leaving all host bits 0 = 255.255.11111110.00000000 = 255.255.254.0. What is the direct broadcast IP address that this workstation should use? 128.249.27.255(2 points). Set all host bits to 1 = 128.249.00011011.11111111 = 128.249.27.255. How many addresses are available for assignment from this subnetwork (include.17)? 2^9 2, or 510(2 points). 23 bits of network (/23) leaves 9 bits for host, 2^9 2. 9. We have found the following CIDR block route entry in our router: 195.10.12.0/23 How many Class C addresses does this summarized route identify? Circle the correct answer. b. 2 (10 points). All or nothing on this one. Class C would normally be /24. We are told this is /23, so on extra bit has been reserved for the customer, thereby combining 2 possible class C s: the one whose 23 rd bit is 0, and the one whose 23 rd bit is 1. Page 8 of 9
10. We wish to examine the following network of PCs, Ethernet segments, and Bridges: S1 S2 S3 S4 S5 S1 S5 LAN1 LAN2 LAN3 B1 B2 port 1 port 2 port 1 port 2 Address Port Address Port S1 1 S1 1 S2 1 S2 1 S5 2 S5 1 S4 2 Show the final state of the B1 and B2 MAC tables as the following events occur. Assume each MAC table has enough memory to hold five entries as shown, implemented as a most-recently-used buffer. The first entry has been shown for you already. 1. S1 sends a packet to S5 (shown on the diagram already) 2. S2 sends a packet to S5 recorded only in right table 3. S5 sends a packet to S3 recorded only in right table 4. S4 sends a packet to S5 recorded only in right table 5. S3 generates a loopback IP packet generates no packet 6. S5 is physically moved to LAN 2 generates no packet 7. S5 sends a packet to S2 updated in both tables (10 points) Getting the S5 entry updated in table two properly was worth 4 points. Page 9 of 9