CS417 Exam 1

Fall 2011

Paul Krzyzanowski

    Part I – 20 points

  1. 4 points
    (a) From an application’s perspective, what are the differences between virtual circuit service and real circuit-switched networking?
    (b) What are the similarities?
  2. 4 points
    Under what conditions does jitter increase when using TCP versus UDP?
  3. 4 points
    The National Institute of Standards and Technology (NIST), an agency of the U.S. Department of Commerce, is responsible for keeping track of the official time in the U.S. Why not just sync your time from a stratum 1 server directly, such as time.nist.gov, and get the time from the most accurate source instead of syncing with a higher-stratum server?
  4. 4 points
    (a) You are synchronizing with a server using Cristian's algorithm. You send a message at 1:10.100 (according to your clock). The server receives the message at 1:15.000 (according to its clock), processes the request, and sends a response containing 1:15.005, which you receive at 1:10.150. To what value do you set your clock?
    (b) If the error of the clock at the server was ±0.015 sec and if the best-case round-trip time to the server is 30 msec. (0.030 sec), what is the time error of the synchronized clock at the client (express as ±E)?
  5. 4 points
    Compare reference count based distributed garbage collection with lease-based garbage collection. What advantages of the latter made it become the dominant technique for distributed garbage collection?
  6. Part II – 60 points – 3 points each

    For each statement, select the most appropriate answer. You may omit one question. Please indicate your choice clearly.

  7. Metcalfe's law isn't really a law but basically states that:
    (a) A network gets more useful with more people on it.
    (b) Processors get twice as powerful every two years.
    (c) Transistors get twice as small every two years.
    (d) Network bandwidth doubles every two years.
  8. A multiprocessor system is best characterized as one where the processors:
    (a) Connect to a data network.
    (b) Communicate over a shared bus.
    (c) Share system memory.
    (d) Share the same cache memory.
  9. An SMP (symmetric multiprocessor) system is a multiprocessor system where:
    (a) The processors engage in the same computation for fault tolerance.
    (b) All processors have equal access to memory.
    (c) An even number (2, 4, …) of processors allows workloads to be partitioned.
    (d) Two or more computers are connected to a shared network.
  10. A snoopy cache reduces traffic on the system bus:
    (a) For read operations.
    (b) For write operations.
    (c) For both read and write operations.
    (d) None of the above; it has no effect on bus traffic.
  11. A single system image refers to:
    (a) multiple computers that look and behave like one system.
    (b) a single operating system that controls a network of computers.
    (c) a cloned set of computers where multiple machines run the same software.
    (d) an instance of a single computer connected to a network.
  12. A snoopy cache is designed for this architecture:
    (a) Bus-based.
    (b) Crossbar switch.
    (c) NUMA.
    (d) All of the above.
  13. TCP/IP occupies which layer of the OSI protocol stack?
    (a) Data Link (2)
    (b) Network (3)
    (c) Transport (4)
    (d) Presentation (6)
  14. UDP/IP occupies which layer of the OSI protocol stack?
    (a) Data Link (2)
    (b) Network (3)
    (c) Transport (4)
    (d) Presentation (6)
  15. Using ARP, the Address Resolution Protocol, how do you find the corresponding ethernet address for a machine's IP address?
    (a) Broadcast a network request containing the IP address, asking for the corresponding IP address.
    (b) Contact an ARP name server that contains IP to ethernet address mappings.
    (c) Use a subset of the bits of the machine's IP address to create the ethernet address.
    (d) That is not the purpose of ARP: ethernet and IP addresses are unrelated.
  16. Differentiated services (e.g., DiffServ) is an example of:
    (a) Packet compression.
    (b) Flow detection.
    (c) Admission control.
    (d) Traffic control.
  17. On a network, jitter is best defined as the variation:
    (a) in bandwidth together with the variation in delay among packets.
    (b) in bandwidth among packets.
    (c) in the packet error rate among packets.
    (d) in delay among packets.
  18. A DNS referral is an example of:
    (a) Static binding.
    (b) Dynamic binding.
    (c) Iterative name resolution.
    (d) Recursive name resolution.
  19. What is the relationship between an IP address and an Internet domain name?
    (a) All hosts within an IP network must be under the same higher-level domain name (e.g., all machines within 128.6.*.* must be under the .rutgers.edu name).
    (b) Each IP address must have a unique IP domain name associated with it.
    (c) Both (a) and (b).
    (d) There is none.
  20. There are three systems on your network: A, B, C. Prior to synchronization, A's clock reads 2:30, B's clock reads: 2:36, and C's clock reads 2:42. B is the master. After synchronization via the Berkeley algorithm, what is the time on A's clock (ignore synchronization or network latencies)?
    (a) 2:30
    (b) 2:33
    (c) 2:36
    (d) 2:39
  21. Based on their vector timestamps, which event causally precedes (4, 2, 8, 5)?
    (a) (3, 1, 7, 7)
    (b) (5, 1, 6, 2)
    (c) (4, 2, 8, 4)
    (d) (4, 3, 8, 5)
  22. The implementation of which message ordering technique is not a feasible one to implement?
    (a) Global time ordering
    (b) Total ordering
    (c) Sync ordering
    (d) Unordered
  23. The following events took place on three systems at the stated physical (time of day) times: System 1: a (3:00), b (3:01) System 2: c (2:55), d (3:04) System 3: e (3:03) Event b is the sending of a message from system 1 to 2. Event d is the receipt of that message. Using Lamport's happened-before relation, which events happened before event d?
    (a) a
    (b) a, b
    (c) a, b, c
    (d) a, d, c, e
  24. IP multicast provides which level of reliability?
    (a) Atomic
    (b) Reliable
    (c) Unreliable
    (d) User-selectable
  25. Sending an IP multicast within an ethernet local area network relies on:
    (a) Sending a message to an IGMP-aware router.
    (b) Converting an IP multicast address to an ethernet multicast address.
    (c) Knowing the IP addresses of members of the group.
    (d) Broadcasting an IP multicast packet.
  26. Microsoft ORPC is essentially DCE RPC with the addition of:
    (a) a multi-canonical data conversion format.
    (b) a unique universal ID.
    (c) an interface pointer.
    (d) a network data representation.
  27. A shared-nothing disk pack is a set of:
    (a) storage servers that provide fault tolerant storage without replication.
    (b) storage servers that provide fault tolerant storage via replication.
    (c) cloned machines that share only the storage in their partition.
    (d) cloned machines that have their own storage.
  28. Part III – 20 points, 2 points each

    For each statement, specify whether it is true or false by circling the correct choice. You may omit one question. Please indicate your choice clearly.

  29. A surrogate process in Microsoft DCOM is a server process that instantiates remote object requests.
    [True] [False]
  30. If event a happened before event b, then clock(a) must be less than clock(b) according to Lamport's algorithm.
    [True] [False]
  31. If clock(a) < clock(b) then event a must have happened before event b according to Lamport's algorithm.
    [True] [False]
  32. A farm is a collection of servers, applications, and data spread across multiple sites.
    [True] [False]
  33. A broadband network is faster than a baseband network over the same type of medium.
    [True] [False]
  34. Class-based IP addressing allows traffic to be divided into different levels of service.
    [True] [False]
  35. An ethernet driver encapsulates an IP packet within an ethernet packet before sending it onto the network.
    [True] [False]
  36. The Real Time Protocol, RTP, provides admission-based quality of service on IP networks.
    [True] [False]
  37. A linear compensating function adjusts for the effects of clock drift.
    [True] [False]
  38. XML-RPC is an example of explicit typing.
    [True] [False]
  39. WSDL (Web Services Description Language) can serve as an Interface Description Language for SOAP.
    [True] [False]
  40. A surrogate process in Microsoft COM+ is a server process that loads COM libraries.
    [True] [False]
Last modified March 24, 2020.
recycled pixels