Computer Networks Final

Packet scheduling (Scenario 1, FCFS). Consider the pattern of red and green packet arrivals to a router's output port queue, shown below. Suppose each packet takes one time slot to be transmitted, and can only begin transmission at the beginning of a time slot after its arrival. Indicate the sequence of departing packet numbers (at t = 1, 2, 3, 4, 5, 7, 8) under FCFS scheduling. Give your answer as 7 ordered digits (each corresponding to the packet number of a departing packet), with a single space between each digit, and no spaces before the first or after the last digit, e.g., in a form like 7 6 5 4 3 2 1).

1 2 3 4 5 6 7

Which of the following addresses can not be used by an interface in the 223.1.3/29 network? Check all that apply.

1. 223.1.3.16 2. 223.1.2.6 3. 223.1.3.28

Subnetting(a). Consider the three subnets in the diagram below. What is the maximum # of interfaces in the 223.1.2/24 network?

256

What is the maximum # of interfaces in the 223.1.3/29 network?

8

Full buffers! Suppose that the output port has enough buffer space for only three packets, including the packet being transmitted (in practice, routers have a lot more buffer space!). What happens to packet 4 in the example above, when it arrives to find the output buffers full?

Packet 4 is dropped or a queued packet is dropped from the buffer to make space for packet 4.

What's inside a router? Match the names of the principal router components (A,B,C,D below) with their function and whether they are in the network-layer data plane or control plane.

(A) are ... input ports, operating primarily in the data plane. (B) is ... the switching fabric, operating primarily in the data plane. (C) are ... output ports, operating primarily in the data plane. (D) is ... the routing processor, operating primarily in the control plane.

Dijkstra's Algorithm (1, part 1). Consider the network shown below, and Dijkstra's link-state algorithm to find the least cost path from source node U to all other destinations. Using the algorithm statement and its visual representation used in the textbook, complete the first row in the table below showing the link state algorithm's execution by matching the table entries (a), (b), (c), and (d) with their values. Write down your final [correct] answer, as you'll need it for the next question.

(a) 2,u (b) 8,u (c) 3,u (d) infinity

How to forward to a border router? Consider the network shown below. Suppose AS1, AS2, AS3, and AS4 are running OSPF for their intra-AS routing protocol and that all links have a weight of 1. Now suppose the link between 2a and 4a is up, and that paths to x via AS2 and AS4 are known with AS1. Hot potato routing is used in conjunction with iBGP within an AS for determining the outgoing border router to use. Indicate which one of the statements below are true.

1d will forward along z2 since 1d has a shorter intra-domain path than border router 1b to border router 1c, and hot potato routing is used.

Dijkstra's link-state routing algorithm (Part 2). Consider the graph shown below and the use of Dijkstra's algorithm to compute a least cost path from a to all destinations. Suppose that nodes b and d have already been added to N'. What is the path cost to the next node to be added to N' (refer to the text or page 12 of lecture slides for an explanation of notation).

4

HOL blocking. What is meant by Head of the Line (HOL) blocking?

A queued datagram waiting for service at the front of a queue prevents other datagrams in queue from moving forward in the queue.

Generalized forwarding. What is meant by generalized forwarding (as opposed to destination-based forwarding) in a router or switch?

Any of several actions (including drop (block), forward to a given interface, or duplicate-and-forward) can be made based on the contents of one or more packet header fields.

IPv4/IPv6 co-existence: tunneling (a). Consider the mixed IPv4/IPv6 network shown below, where an IPv4 tunnel exists between IPv6 routers B and E. Suppose that IPv6 router A sends a datagram to IPv6 router F. IPv6 datagrams are shown in blue; the IPv4 datagram is in red (containing the encapsulated IPv6 datagram in blue). Perform the matching below to indicate the datagram field value and type at point (a).

At point (a), the IP version field in the datagram is: IPv6 At point (a), the source IP address is that of host: A At point (a), the destination IP address is that of host: F At point (a), the number of bits in the destination IP address is: 128

IPv4/IPv6 co-existence: tunneling (a). Consider the mixed IPv4/IPv6 network shown below, where an IPv4 tunnel exists between IPv6 routers B and F. Suppose that IPv6 Host 2 sends a datagram to IPv6 Host 4; the sequence of routers traversed are B,c,b,F. IPv6 datagrams are shown in blue; the IPv4 datagram is in red (containing the encapsulated IPv6 datagram in blue). Perform the matching below to indicate the datagram field value and type at point (a).

At point (a), the IP version field in the datagram is: IPv6 At point (a), the source IP address is that of: host 2 At point (a), the destination IP address is that of: host 4 At point (a), the number of bits in the destination IP address is: 128

Perform the matching below to indicate the datagram field value and type at point (b).

At point (b), the larger (red, encapsulating) datagram has an IP version field in the datagram of: IPv4 At point (b), the larger (red, encapsulating) datagram has a source IP address of:  router B At point (b), the larger (red, encapsulating) datagram has a destination IP address of:  router F At point (b), the smaller (blue, encapsulated) datagram has a source IP address of:  host 2 At point (b), the smaller (blue, encapsulated) datagram has an IP version field in the datagram of: IPv6

Perform the matching below to indicate the datagram field value and type at point (c).

At point (c), the IP version field in the datagram is: IPv6 At point (c), the source IP address is that of: host 2 At point (c), the destination IP address is that of: host 4 At point (c), the number of bits in the destination IP address is: 128

Where does "match+action" happen? Where in a router does "match plus action" happen to determine the appropriate output port to which the arriving datagram should be directed?

At the input port where a packet arrives.

Where does destination address lookup happen? Where in a router is the destination IP address looked up in a forwarding table to determine the appropriate output port to which the datagram should be directed?

At the input port where a packet arrives.

BGP advertisement policy. Consider the network below, and assume that a provider network only wants to carry traffic to or from its customer networks. Which of the following statements are true?

B will advertise a route Bx to A, C and D, since A, C and D need to know how to route to B's customer network x.

Bellman Ford Algorithm (2, part 1). Consider the scenario shown below, where at t=1, node e receives distance vectors (DVs) from neighboring nodes b, d, and f. The (old) DV at e (before receiving the new DVs from its neighbors) is also shown, as well as the DVs being sent from b, d, and f. Select the new distance vector components at e, below by matching one of e's new DV entries to its value.

De(b) 4 De(d) 1 De(e) 0 De(f) 5 De(h) 3

Network Address Translation (NAT). Which one of the following operations is not performed by NAT.?

Generating ACKs back to the TCP sender and then taking responsibility for reliably delivery the segment to its destination, possibly using a non-TCP reliable data transfer protocol.

eBGP or iBGP? Consider routers 2c and 2d in Autonomous System AS2 in the figure below. Indicate the flavor of BGP and the router from which each of 2c and 2d learns about the path to destination x.

How does router 2c learn of the path AS3, X to destination network X? From 3a via eBGP. How does router 2d learn of the path AS3, X to destination network X? From 2c via iBGP.

Internal structure of the SDN controller (3). Which of the functions below belong in the controller layer labeled "Communication to/from controlled device"? Check all below that apply.

OpenFlow protocol

Path advertisement and policy (Part 2). Again, suppose a provider network only wants to carry traffic to/from its customer networks (i.e., to provide no transit service), and customer networks only want to carry traffic to/from itself. Suppose C has advertised path Cy to A. To implement this policy, to which of the following networks would network A advertise the path ACy?

w

BGP advertisement policy. Suppose that a network that is a customer of two different provider networks will not relay traffic between its two provider networks. How can a customer network such as x in the figure below implement that policy.

x will not advertise to provider networks B or C that it has a path to the other provider network

Dijkstra's Algorithm (2, part 1). Consider the network shown below, and Dijkstra's link-state algorithm. Here, we are interested in computing the least cost path from node y (note: not node u!) to all other nodes using Dijkstra's algorithm. Using the algorithm statement used in the textbook and its visual representation, complete the first row in the table below showing the link state algorithm's execution by matching the table entries (a), (b), (c), and (d) with their values. Write down your final [correct] answer, as you'll need it for the next question.

(a) infinity (b) 1,y (c) 6,y (d) 8,y

Dijkstra's Algorithm (1, part 2). Consider the network shown below, and Dijkstra's link-state algorithm to find the least cost path from source node U to all other destinations. Using the algorithm statement and its visual representation used in the textbook, complete the second row in the table below showing the link state algorithm's execution by matching the table entries (a), (b), (c), (d) and (e) with their values. Write down your final [correct] answer, as you'll need it for the next question; the *s shown correspond to your answers to the question 5.01-1.

(a) uv (b) 2,u (c) 6,v (d) 3,u (e) infinity

Dijkstra's Algorithm (1, part 3). Consider the network shown below, and Dijkstra's link-state algorithm to find the least cost path from source node U to all other destinations. Using the algorithm statement and its visual representation used in the textbook,complete the third row in the table below showing the link state algorithm's execution by matching the table entries (a), (b), (c), (d) and (e) with their values. Write down your final [correct] answer, as you'll need it for the next question; the *s shown correspond to your answers to earlier parts of this question.

(a) uvx (b) 2,u (c) 5,x (d) 3,u (e) 9,x

Dijkstra's Algorithm (1, part 4). Consider the network shown below, and Dijkstra's link-state algorithm to find the least cost path from source node U to all other destinations. Using the algorithm statement and its visual representation used in the textbook, complete the fourth row in the table below showing the link state algorithm's execution by matching the table entries (a), (b), (c), and (d) with their values. Write down your final [correct] answer, as you'll need it for the next question. The *s shown correspond to your answers to the earlier parts of this question; note that a couple of table entries are given for you (!).

(a) uvxw (b) 5,x (c) 8,w (d) 6,w

Dijkstra's Algorithm (1, part 5). Consider the network shown below, and Dijkstra's link-state algorithm to find the least cost path from source node U to all other destinations. Using the algorithm statement and its visual representation used in the textbook, complete the fifth row in the table below showing the link state algorithm's execution by matching the table entries (a), (b), (c),and (d) with their values. The *s shown correspond to your earlier answers to this question.

(a) uvxwz (b) 5,x (c) 7,z (d) 6,w

Dijkstra's Algorithm (2, part 2). Consider the network shown below, and Dijkstra's link-state algorithm. Here, we are interested in computing the least cost path from node y (note: not node u!) to all other nodes using Dijkstra's algorithm. Using the algorithm statement used in the textbook and its visual representation, complete the first row in the table below showing the link state algorithm's execution by matching the table entries (a), (b), (c), (d) and (e) with their values. Write down your final [correct] answer, as you'll need it for the next question.

(a) yw (b) 9,w (c) 6,w (d) 3,w (e) 5,w

Dijkstra's Algorithm (2, part 3). Consider the network shown below, and Dijkstra's link-state algorithm. Here, we are interested in computing the least cost path from node y (note: not node u!) to all other nodes using Dijkstra's algorithm. Using the algorithm statement used in the textbook and its visual representation, complete the first row in the table below showing the link state algorithm's execution by matching the table entries (a), (b), (c), (d), and (e) with their values. Write down your final [correct] answer, as you'll need it for the next question.

(a) ywx (b) 6,x (c) 5,x (d) 3,w (e) 4,x

Dijkstra's Algorithm (2, part 4). Consider the network shown below, and Dijkstra's link-state algorithm. Here, we are interested in computing the least cost path from node y (note: not node u!) to all other nodes using Dijkstra's algorithm. Using the algorithm statement used in the textbook and its visual representation, complete the first row in the table below showing the link state algorithm's execution by matching the table entries (a), (b), (c), and (d) with their values.

(a) ywxz (b) 6,x (c) 5,x (d) 4,x

Destination-based match+action. Destination-based forwarding, which we studied in section 4.2, is a specific instance of match+action and generalized forwarding. Select the phrase below which best completes the following sentence: "In destination-based forwarding, ..."

... after matching on the destination IP address in the datagram header, the action taken is to forward the datagram to the output port associated with that destination IP address.

Packet scheduling (Scenario 1, Priority). Consider the pattern of red and green packet arrivals to a router's output port queue, shown below. Suppose each packet takes one time slot to be transmitted, and can only begin transmission at the beginning of a time slot after its arrival. Indicate the sequence of departing packet numbers (at t = 1, 2, 3, 4, 5, 7, 8) under priority scheduling, where red packets have higher priority. Give your answer as 7 ordered digits (each corresponding to the packet number of a departing packet), with a single space between each digit, and no spaces before the first or after the last digit, e.g., in a form like 7 6 5 4 3 2 1).

1 2 3 5 4 7 6

Packet scheduling (c). Consider the same pattern of red and green packet arrivals to a router's output port queue, shown below. Suppose each packet takes one time slot to be transmitted, and can only begin transmission at the beginning of a time slot after its arrival. Indicate the sequence of departing packet numbers (at t = 1, 2, 3, 4, 5, 6, 7) under round robin scheduling. Assume a round-robin scheduling cycle begins with green packets. Give your answer as 7 ordered digits (each corresponding to the packet number of a departing packet), with a single space between each digit, and no spaces before the first or after the last digit, e.g., in a form like 7 6 5 4 3 2 1).

1 2 4 3 6 5 7

Generalized match+action. Which of the following match+actions can be taken in the generalized OpenFlow 1.0 match+action paradigm that we studied in Section 4.4? Check all that apply.

1. ... after matching on the 48-bit link-layer destination MAC address, the action taken is to forward the datagram to the output port associated with that link-layer address. 2. ... after matching on the source and destination IP address in the datagram header, the action taken is to forward the datagram to the output port associated with that source and destination IP address pair. 3. ... after matching on the port number in the segment's header, the action taken is to decide whether or not to drop that datagram containing that segment. 4. ... after matching on the destination IP address in the datagram header, the action taken is to forward the datagram to the output port associated with that destination IP address. 5. ... after matching on the destination IP address in the datagram header, the action taken is to decide whether or not to drop that datagram. 6. ... after matching on the port number in the segment's header, the action taken is to forward the datagram to the output port associated with that destination IP address.

IPv4 versus IPv6. Which of the following fields occur ONLY in the IPv6 datagram header (i.e., appear in the IPv6 header but not in the IPv4 header)? Check all that apply.

1. 128-bit source and destination IP addresses. 2. The flow label field.

Two dimensional parity. Which of the following statements is true about a two-dimensional parity check (2D-parity) computed over a payload?

1. 2D-parity can detect any case of a single bit flip in the payload. 2. 2D-parity can detect and correct any case of a single bit flip in the payload. 3. 2D-parity can detect any case of two bit flips in the payload.

Path advertisement and policy (Part 1). Suppose a provider network only wants to carry traffic to/from its customer networks (i.e., to provide no transit service), and customer networks only want to carry traffic to/from itself. Consider the figure below. To implement this policy, to which of the following networks would network C advertise the path Cy?

1. A 2. x 3. B

DHCP request message. Which of the following statements about a DHCP request message are true (check all that are true). Hint: check out Figure 4.24 in the 7th and 8th edition of our textbook.

1. A DHCP request message is sent broadcast, using the 255.255.255.255 IP destination address. 2. A DHCP request message may contain the IP address that the client will use. 3. The transaction ID in a DHCP request message will be used to associate this message with future DHCP messages sent from, or to, this client.

Forwarding versus routing. Consider the travel analogy discussed in the textbook - some actions we take on a trip correspond to forwarding and other actions we take on a trip correspond to routing. Which of the following travel actions below correspond to forwarding? The other travel actions that you don't select below then correspond to routing.

1. A car takes the 3rd exit from a roundabout. 2. A car waits at light and then turns left at the intersection. 3. A car stops at an intersection to "gas-up" and take a "bathroom break"

What type of control plane? We've seen that there are two approaches towards implementing the network control plane - a per-router control-plane approach and a software-defined networking (SDN) control-plane approach. Which of the following actions occur in a per-router control-plane approach? The other actions that you don't select below then correspond to actions in an SDN control plane.

1. A router exchanges messages with another router, indicating the cost for it (the sending router) to reach a destination host. 2. Routers send information about their incoming and outgoing links to other routers in the network.

What is a subnet? What is meant by an IP subnet? (Check zero, one or more of the following characteristics of an IP subnet).

1. A set of device interfaces that can physically reach each other without passing through an intervening router. 2. A set of devices that have a common set of leading high order bits in their IP address.

What is an IP address actually associated with? Which of the following statements is true regarding an IP address? (Zero, one or more of the following statements is true).

1. An IP address is associated with an interface. 2. If a host has more than one interface, then it has more than one IP address at which it can be reached. 3. If a router has more than one interface, then it has more than one IP address at which it can be reached.

Bellman Ford Algorithm (1, part 2). Consider the scenario shown below, where at t=1, node e receives distance vectors from neighboring nodes d, b, h and f. The (old) distance vector at e (the node at the center of the network) is also shown, before receiving the new distance vector from its neighbors. Indicate which of the components of new distance vector at e below have a value of 2 after e has received the distance vectors from its neighbors and updated its own distance vector.

1. De(c) 2. De(i) 3. De(g) 4. De(a)

Bellman Ford Algorithm (1, part 1). Consider the scenario shown below, where at t=1, node e receives distance vectors from neighboring nodes d, b, h and f. The (old) distance vector at e (the node at the center of the network) is also shown, before receiving the new distance vector from its neighbors. Indicate which of the components of new distance vector at e below have a value of 1 after e has received the distance vectors from its neighbors and updated its own distance vector.

1. De(d) 2. De(h) 3. De(b) 4. De(f)

Routing versus forwarding. Which of the following statements correctly identify the differences between routing and forwarding. Select one or more statements.

1. Forwarding refers to moving packets from a router's input to appropriate router output, and is implemented in the data plane.

The "thin waist" of the Internet. What protocol (or protocols) constitutes the "thin waist" of the Internet protocol stack? Check all that apply.

1. IP

What fields can be matched in generalized match+action. Which of the following fields in the frame/datagram/segment/application-layer message can be matched in OpenFlow 1.0? Check all that apply.

1. IP type-of-service field 2. IP source address 3. Upper layer protocol field 4. IP destination address 5. Source and/or destination port number

SDN implementation of Dijkstra's algorithm. Consider the implementation of Dijkstra's algorithm in an SDN framework. Which of the following statements are true? (Hint: more than one statement is true.)

1. If a router's forwarding table should be changed as a result of running Dijkstra's algorithm, the new flow table for that router will be updated by the SDN controller via the southbound API using the Openflow protocol. 2. When executing, Dijkstra's algorithm will run as a network control application "on top" on the SDN controller. 3. When executing, Dijkstra's algorithm will use the link-state database that is maintained within the SDN controller.

Crafting network-wide forwarding using flow tables. Consider the network below. We want to specify the match+action rules at s3 so that only the following network-wide behavior is allowed: traffic from 128.119/16 and destined to 137.220/16 is forwarded on the direct link from s3 to s1; traffic from 128.119/16 and destined to 67.56/16 is forwarded on the direct link from s3 to s2; incoming traffic via port 2 or 3, and destined to 128.119/16 is forwarded to 128.119/16 via local port 1. No other forwarding should be allowed. In particular s3 should not forward traffic arriving from 137.220/16 and destined for 67.56/16 and vice versa. From the list of match+action rules below, select the rules to include in s3's flow table to implement this forwarding behavior. Assume that if a packet arrives and finds no matching rule, it is dropped.

1. Input port: 1; Dest: 67.56/16 Action: forward(3) 2. Input port:1 ; Dest: 137.220/16 Action: forward(2) 3. Input port: 3; Dest: 128.119/16 Action: forward(1) 4. Input port: 2; Dest: 128.119/16 Action: forward(1)

Crafting network-wide forwarding using flow tables (more). Consider the network below. We want to specify the match+action rules at s3 so that s3 acts only as a relay for traffic between 137.220/16 and 67.56/16. In particular s3 should not accept/forward and traffic to/from 128.119/16. From the list of match+action rules below, select the rules to include in s3's flow table to implement this forwarding behavior. Assume that if a packet arrives and finds no matching rule, it is dropped.

1. Input port: 3; Dest: 137.220/16 Action: forward(2) 2. Input port: 2; Dest: 67.56/16 Action: forward(3)

Internal structure of the SDN controller (1). Which of the functions below belong in the controller layer labeled "Interface, abstractions for network control apps"? Check all below that apply.

1. Intent 2. Network graph

Open Shortest Path First (OSPF). Consider the OSPF routing protocol. Which of the following characteristics are associated with OSPF (as opposed to BGP)?

1. Is an intra-domain routing protocol. 2. Finds a least cost path from source to destination. 3. Floods link state control information.

Internal structure of the SDN controller (2). Which of the functions below belong in the controller layer labeled "Network-wide distributed, robust state management"? Check all below that apply.

1. Link-state information 2. Host information 3. Switch information 4. Flow tables 5. Statistics

The control plane versus the data plane. For each of the actions below, select those actions below that are primarily in the network-layer data plane. The other actions that you don't select below then correspond to control-plane actions.

1. Looking up address bits in an arriving datagram header in the forwarding table. 2. Moving an arriving datagram from a router's input port to output port 3. Dropping a datagram due to a congested (full) output buffer.

What's a "middlebox"? Which of the following network devices can be thought of as a "middlebox"? Check all that apply.

1. Network Address Translation box 2. HTTP load balancer 3. HTTP cache

Routing within networks? Among the following protocols, terminology or considerations, indicate those that are associated with "routing within a single network (typically owned and operated by one organization)."

1. OSPF 2. intra-AS routing 3. intra-domain routing 4. Driven more by performance than by routing policy.

Open Shortest Path First (OSPF). Check the one or more of the following statements about the OSPF protocol that are true.

1. OSPF implements hierarchical routing 2. OSPF is an intra-domain routing protocol. 3. OSFP uses a Dijkstra-like algorithm to implement least cost path routing.

Pure Aloha and CSMA. Which of the following statements is true about both Pure Aloha, and CSMA (both with and without collision detection?

1. Pure Aloha and CSMA can achieve 100% utilization, in the case that there is only one node that always has frames to send

What is the Internet Protocol? What are the principal components of the IPv4 protocol (check all that apply)?

1. Pv4 datagram format. 2. IPv4 addressing conventions. 3. Packet handling conventions at routers (e.g., segmentation/reassembly)

Link-layer services. Which of the following services may be implemented in a link-layer protocol? Select one or more statements.

1. Reliable data transfer between directly connected nodes. 2. Flow control between directly connected nodes. 3. Coordinated access to a shared physical medium. 4. Bit-level error detection and correction. 5. Multiplexing down from / multiplexing up to a network-layer protocol.

Characteristics of Multiple Access Protocols (a).Consider the following multiple access protocols that we've studied: (1) TDMA, and FDMA (2) CSMA (3) Aloha, and (4) polling. Which of these protocols are collision-free (e.g., collisions will never happen)?

1. TDMA and FDMA 2. Polling

Characteristics of Multiple Access Protocols (b).Consider the following multiple access protocols that we've studied: (1) TDMA, and FDMA (2) CSMA (3) Aloha, and (4) polling. Which of these protocols requires some form of centralized control to mediate channel access?

1. TDMA and FDMA 2. Polling

Characteristics of Multiple Access Protocols (c).Consider the following multiple access protocols that we've studied: (1) TDMA, and FDMA (2) CSMA (3) Aloha, and (4) polling. For which of these protocols is the maximum channel utilization 1 (or very close to 1)?

1. TDMA and FDMA 2. Polling

Characteristics of Multiple Access Protocols (d).Consider the following multiple access protocols that we've studied: (1) TDMA, and FDMA (2) CSMA (3) Aloha, and (4) polling. For which of these protocols is there a maximum amount of time that a node knows that it will have to wait until it can successfully gain access to the channel?

1. TDMA and FDMA 2. Polling

The end-to-end principle. Which of the statements below are true statements regarding the "end-to-end principle"? Check all that apply.

1. The end-to-end argument advocates placing functionality at the network edge because some functionality cannot be completely and correctly implemented in the network, and so needs to be placed at the edge in any case, making in-network implementation redundant. 2. The end-to-end argument allows that some redundant functionality might be placed both in-network and at the network edge in order to enhance performance.

The network layer - where is it? Check all of the statements below about where (in the network) the network layer is implemented that are true.

1. The network layer is implemented in hosts at the network's edge. 2. The network layer is implemented in routers in the network core

Channel partitioning protocols. Which of the following statements is true about channel partitioning protocols?

1. There can be times when the channel is idle, when a node has a frame to send, but is prevented from doing so by the medium access protocol. 2. Channel partitioning protocols can achieve 100% channel utilization, in the case that all nodes always have frames to send.

Polling and token-passing protocols. Which of the following statements is true about polling and token-passing protocols?

1. These protocol can achieve close to 100% channel utilization, in the case that all nodes always have frames to send (the fact that the utilization is close to, but not exactly, 100% is due to a small amount of medium access overhead but not due to collisions) 2. These protocol can achieve close 100% utilization, in the case that there is only one node that always has frames to send (the fact that the utilization is close to, but not exactly, 100% is due to a small amount of medium access overhead but not due to collisions)

Different types of addressing (a). We've now learned about both IPv4 addresses and MAC addresses. Consider the address properties below, and use the pulldown menu to indicate which of these properties is only a property of MAC addresses (and therefore is not a property of IPv4 addresses - careful!).

1. This is a 48-bit address. 2. This is a link-layer address.

Longest prefix matching. Consider the following forwarding table below. Indicate the output to link interface to which a datagram with the destination addresses below will be forwarded under longest prefix matching. (Note: The list of addresses is ordered below. If two addresses map to the same output link interface, map the first of these two addresses to the first instance of that link interface.)

11001000 00010111 00010010 10101101 This is the first destination address in the list that maps to output port 0. 11001000 00010111 00011000 00001101 This is the first destination address in the list that maps to output port 1. 11001000 00010111 00011001 11001101 This is the first destination address in the list that maps to output port 2. 10001000 11100000 00011000 00001101 This is the first destination address in the list that maps to output port 3. 11001000 00010111 00011000 11001111 This is the second destination address in the list that maps to output port 1. 11001000 00010111 00010001 01010101 This is the second destination address in the list that maps to output port 0. 11001000 00010111 00011101 01101101 This is the second destination address in the list that maps to output port 2.

How to forward to a border router? Consider the network shown below. Suppose AS1, AS2, AS3, and AS4 are running OSPF for their intra-AS routing protocol and that all links have a weight of 1. Initially suppose there is no link between AS2 and AS4. Once router 1d learns about destination x, it will need to install a forwarding table entry to x. Indicate which one of the statements below are true.

1d will forward along z1 since OSPF has computed the path to 1c is via z1.

Perform the matching below to indicate the datagram field value and type at point (b).

At point (b), the larger (red, encapsulating) datagram an IP version field in the datagram of: IPv4 At point (b), the larger (red, encapsulating) datagram has a source IP address of host:  B At point (b), the larger (red, encapsulating) datagram has a destination IP address of host:  E At point (b), the smaller (blue, encapsulated) datagram has a source IP address of host:  A At point (b), the smaller (blue, encapsulated) datagram has a destination IP address version of:  IPv6

What type of routing? Match the name of a general approach to routing with characteristics of that approach.

Centralized, global routing. All routers have complete topology, and link cost information. Decentralized routing. An iterative process of computation, exchange of informatoin with neighbors. Routers may initially only know link costs to directly-attached neighbors. Static routing. Routes change slowly over time. Dynamic routing. Routing changes quickly over time.

Best effort service. Which of the following quality-of-service guarantees are part of the Internet's best-effort service model? Check all that apply.

None of the other services listed here are part of the best-effort service model. Evidently, best-effort service really means no guarantees at all!

(a) Suppose we want to implement the following rule: r2 should act as a firewall, only allowing TCP traffic into the 22.33/16 network from any network. Specify a single flow table row entry to implement this rule, indicating the column entries for the row below. The * is a wildcard match, which matches everything.

In the "source IP" column, the flow table entry should be: * In the "dest. IP" column, the flow table entry should be: 22.33/16 In the "protocol" column, the flow table entry should be: TCP In the "action" column, the flow table entry should be: forward(3)

Path advertisement and policy (Part 3). Again, suppose a provider network only wants to carry traffic to/from its customer networks (i.e., to provide no transit service), and customer networks only want to carry traffic to/from itself. Suppose C has advertised path Cy to x. To implement this policy, to which of the following networks would network x advertise the path xCy?

None of these other networks

(b) Suppose we want to implement the following behavior, and that the default configuration is that all traffic from 22.33/16 should be allowed to be forwarded by r2 to the outside network. However, we also want to implement a higher priority rule (which would be earlier in the table, for example) so that users within the 22.33/16 network are never allowed to connect to an external email server on port 25. Specify the flow table row entries to implement this rule, by indicating the column entries below. The * is a wildcard match, which matches everything - you should use * over more specific answers whenever possible.

In the "source IP" column, the flow table entry should be: 22.33/16 In the "dest. IP" column, the flow table entry should be: * In the "dest port" column, the flow table entry should be: 25 In the "action" column, the flow table entry should be: drop

(b) Suppose we want to implement the following behavior, and that the default configuration is that all traffic from 22.33/16 should be allowed to be forwarded by r2 to the outside network. However, we also want to implement a higher priority rule (which would be earlier in the table, for example) so that users within the 22.33/16 network are never allowed to connect to an external web server on port 80. Specify the flow table row entries to implement this rule, by indicating the column entries below. The * is a wildcard match, which matches everything - you should use * over more specific answers whenever possible.

In the "source IP" column, the flow table entry should be: 22.33/16 In the "dest. IP" column, the flow table entry should be: * In the "dest port" column, the flow table entry should be: 80 In the "action" column, the flow table entry should be: drop

(c) Suppose we want to implement least-cost-path forwarding, so that packets from source network 22.33/16 to destination networks 128.119/16 and 53.106/16 are forwarded on the direct link from r2 to each of these networks. Specify the flow table rule to implement this rule for forwarding to destination 128.116/16, by indicating the column entries below. [Hint: you should not assume that datagrams arriving to r2 on interface 3 necessarily have a source IP address of 22.33/16, since as we learned it is easy to "spoof" an IP addresses; so your rule should specify that only datagrams with a source IP address in 22.33.16 should be forwarded out of the network - a process known as "egress filtering".) The * is a wildcard match, which matches everything - you should use * over more specific answers whenever possible.

In the "source IP" column, the flow table entry should be: 22.33/16 In the "dest. IP" column, the flow table entry should be: 128.119/16 In the "dest port" column, the flow table entry should be: * In the "action" column, the flow table entry should be: forward(2)

Routing within or among networks. Match the terms "interdomain routing" and intradomain routing" with their definitions. Recall that in Internet parlance, an "AS" refers to "Autonomous System" - a network under the control of a single organization.

Interdomain routing. Routing among different ASes ("networks"). Intradomain routing. Routing among routers within same AS ("network").

Plug-and-play. What is meant by saying that DHCP is a "plug and play" protocol?

No manual configuration is needed for the host to join the network.

Consider the scenario shown below. The figure below shows the (old) DV at e (before receiving the new DVs from its neighbors) as well as the DVs being sent from b, d, and f. In the previous quesiton you computed the new DV at ate. Now suppose that all network nodes have iterated and completed all of the DV calculations, i.e, that the algorithm has converged and quiesced. Suppose now that sometime after the algorithm has converged, the link between e and f goes down. Will node e send out a new DV to its neighbors? Pick a response below that best answers this question.

No. Node e's distance vector does not change when the link between e and f goes down (since e's shortest path to f did not use this direct link between e and f), so e will not send out a new DV.

Approaches towards implementing the control plane. Match the name of the approach towards implementing a control plane with a description of how this approach works.

Per-router control plane. Individual routing algorithm components - with a component operating in each and every router - interact with each other in the control plane. The individual routing algorithm component executing in a given router computes the local forwarding table. Software-defined networking (SDN). A (typically) remote controller gathers information from routers, and then computes and installs the forwarding tables in routers.

eBGP, iBGP, OSPF? Consider the network shown below. Suppose AS1, AS2, AS3, and AS4 are running OSPF for their intra-AS routing protocol. Suppose eBGP and iBGP are used for the inter-AS routing protocol. Initially suppose there is no link between AS2 and AS4. Indicate the protocol by which a router learns about the network prefix x, which is attached to AS4.

Router 1c learns about prefix x from which protocol? eBGP Router 1d learns about prefix x from which protocol? iBGP

eBGP, iBGP, OSPF? Consider the network shown below. Suppose AS1, AS2, AS3, and AS4 are running OSPF for their intra-AS routing protocol. Suppose eBGP and iBGP are used for the inter-AS routing protocol. Initially suppose there is no link between AS2 and AS4. Indicate the protocol by which a router learns about the network prefix x, which is attached to AS4.

Router 3c learns about prefix x from which protocol? eBGP Router 3a learns about prefix x from which protocol? iBGP

What's a "good" path? What is the definition of a "good" path for a routing protocol? Chose the best single answer.

Routing algorithms typically work with abstract link weights that could represent any of, or combinations of, all of the other answers.

SDN Layers. Consider the SDN layering shown below. Match each layer name below with a layer label (a), (b) or (c) as shown in the diagram.

SDN Controller (network operating system)(b) SDN-controlled switches (c) Network-control applications (a)

Match+action in Openflow 1.0. Consider the figure below that shows the generalized forwarding table in a router. Recall that a * represents a wildcard value. Now consider an arriving datagram with the IP source and destination address fields indicated below. For each source/destination IP address pair, indicate which rule is matched. Note: assume that a rule that is earlier in the table takes priority over a rule that is later in the table and that a datagram that matches none of the table entries is dropped.

Source: 1.2.56.32 Destination:128.116.40.186 Rule 2, with action drop Source: 65.92.15.27 Destination: 3.4.65.76 Rule 1, with action forward(2) Source: 10.1.2.3 Destination: 7.8.9.2 Rule 3, with action send to controller Source: 10.1.34.56 Destination: 54.72.29.90 Nomatch to any rule.

The Internet hourglass. What is meant when it is said that the Internet has an "hourglass" architecture? See the picture below if you are unfamiliar with an "hourglass".

The Internet protocol stack has a "thin waist" in the middle, like an hourglass. The Internet Protocol (IP) is the only network-layer protocol in the middle layer of the stack. Every other layer has multiple protocols at that layer.

Packet dropping. Suppose a datagram is switched through the switching fabric and arrives to its appropriate output to find that there are no free buffers. In this case:

The packet will either be dropped or another packet will be removed (lost) from the buffer to make room for this packet, depending on policy. But the packet will definitely not be be sent back to the input port.

Network Address Translation (a). Consider the following scenario in which host 10.0.0.1 is communicating with an external web server at IP address 128.119.40.186. The NAT table shows the table entry associated with this TCP flow. What are the source and destination IP address and port numbers at point A?

The source IP address is: 10.0.0.1 The destination IP address is: 128.119.40.186 The source port number is: 2020 The destination port number is: 80

Network Address Translation (b). Consider the following scenario in which host 10.0.0.1 is communicating with an external web server at IP address 128.119.40.186. The NAT table shows the table entry associated with this TCP flow. What are the source and destination IP address and port numbers at point B?

The source IP address is: 138.76.29.7 The destination IP address is: 128.119.40.186 The source port number is: 5051 The destination port number is: 80

Purpose of DHCP. What is the purpose of the Dynamic Host Configuration Protocol?

To obtain an IP address for a host attaching to an IP network.

The IPv4 header. Match each of the following fields in the IP header with its description, function or use.

Version field This field contains the IP protocol version number. Type-of-service field This field contains ECN and differentiated service bits. Fragmentation offset field This field is used for datagram fragmentation/reassembly. Time-to-live field The value in this field is decremented at each router; when it reaches zero, the packet must be dropped. Header checksum field This field contains the Internet checksum of this datagram's header fields. Upper layer field This field contains the "protocol number" for the transport-layer protocol to which this datagram's payload will be demultiplexed - UDP or TCP, for example. Payload/data field This field contains a UDP or TCP segment, for example. Datagram length field. This field indicates the total number of bytes in datagram.

Subnet addressing. Consider the three subnets below, each in the larger 128.119.160/24 network. The following questions are concerned with subnet addressing. Answer each question by selecting a matching answer. Each answer can be used to answer only one question.

What is the maximum number of hosts possible in the larger 128.119.160/24 network? 256 How many bits are needed to be able to address all of the host in subnet A? 6 Suppose that subnet A has a CIDRized subnet address range of 128.119.160.128/26 (hint: 128 is 1000 0000 in binary); Subnet B has an CIDRied subnet address range of 128.119.160.64/26. We now want a valid CIDRized IP subnet address range for subnet C of the form 128.119.160.x/26. What is a valid value of x? 0

Bellman Ford Algorithm - a change in DV (1, part 4). Consider the network below, and suppose that at t=0, the link between nodes b and c goes down. And so at t=0, node b recomputes its distance vector (DV) and sends out its new DV (as needed). At t=1 this new DV is received at b's neighbors, who then perform their calculation and send out their new DVs (as needed); these new DVs arrive at their neighbors at t=2, and so on. What is the last time in this network at which a DV calculation will take place as a result of the link change at t=0?

an essentially infinite amount of time; this is the count-to-infinity problem

Dijkstra's link-state routing algorithm (Part 1). Consider the graph shown below and the use of Dijkstra's algorithm to compute a least cost path from a to all destinations. Suppose that nodes b and d have already been added to N'. What is the next node to be added to N' (refer to the text or page 12 of lecture slides for an explanation of notation).

e

Bellman Ford Algorithm - a change in DV (1, part 3). Consider the network below, and suppose that at t=0, the link between nodes g and h goes down. And so at t=0, nodes g and h recompute their DVs. Following this recomputation, to which nodes will h send its new distance vector? (Note: to answer this question, you'll need to know some of the DV entries at g and h at t=0, but hopefully they'll be obvious by inspection).

node h does not send out its distance vector, since none of the least costs have changed to any destination.