分布式系统原理与范型考试 2010 答案

分布式系统 2010春季学期期末考试北京大学计算机系，2010年6月7日院系：学号：姓名：一、概念题（共15分）1.asynchronous RPC (3pt) (chap4)With asynchronous RPCs, the server immediately sends a reply back to the client the moment the PRC request is received, after which it calls the requested procedure. The reply acts as an acknowledgment to the client that the server is going to process the RPC. The client will continue without further blocking as soon as it has received the server’s acknowledgment.2.response failure, sate transition failure (3pt) (chap8)The server’s response is simply incorrect. Two kinds of response failures may happen. In the case of a value failure, a server simply provides the wrong reply to a request. The other type of response failure is known as a state transition failure. This kinds of failure happens when the server reacts unexpectedly to an incoming request.3.stateless server, soft state (3pt) (chap3)A stateless server does not keep information on the state of its clients, and can change its own state without having to inform any client.A particular form of a stateless design is where the server maintains what is known as soft sate. The server promises to maintain state on behalf of the client, but only for a limited time. After that time has expired, the server falls back to default behavior.4.totally-ordered multicast, causally-ordered multicasting (3pt) (chap6)A multicast operation by which all messages are delivered in the same order to each receiver.If two messages are not in any way related to each other, we do not care in which order they are delivered to applications. They may even be delivered in different order at different locations.5.Monotonic-read consistency (3pt) (chap7)If a process reads the value of a data item x, any successive read operation on x by that process will always return that same value or a more recent value. In other words, monotonic-read consistency guarantees that if a process has seen a value of x at time t, it will never see and older version of x at a later time.二、简答题（共35分）1. Q: 1) Resolve the following key lookups for the shown Chord-based P2P system: (5pt) (chap5)15@4: 14–18; 22@4: 20–21–28; 30@21: 28–1; 27@21: 28; 18@20: 4–14–182) Adjust the finger tables of nodes 18 and 14 when a node with ID 24 enters the ring. Also give the finger table of node 24. (5pt) (chap5)Node 18: [20,20,24,28,4]; Node 14: [18,18,18,24,1]; Node 24: [28,28,28,1,9].3) Chord allows keys to be looked up recursively or iteratively. Explain the differences, as well as the main advantage of iterative over recursive lookup. (5pt) (chap5)With recursive lookups, a message is forwarded from peer to peer until it reaches its destination. In contrast, with an iterative lookup, the requester is returned the next peer it should ask for the key. One can argue that in the case of Chord, iterative lookups are much better: recursive lookups do not have the advantage of proximity-awareness. Also, note that iterative lookups have the advantage of letting the client handle failures more easily.2. Q: What is a three-tiered client-server architecture? (5pt) (chap 2)A three-tiered client-server architecture consists of three logical layers, where each layer is, in principle, implemented at a separate machine. The highest layer consists of a client user interface, the middle layer contains the actual application, and the lowest layer implements the data that are being used.3. Q: Where does the need for at-least-once and at-most-once semantics come from? Why can’t we have exactly-once semantics? (5pt)The problem originates from having a (suspected) server crash, detected by the lack of a response in the case of an RPC. What the client-side software can do is either resend the request until it finally gets a response (at-least-once semantics) or immediately reports the failure to the client application, thus providing at-most-once semantics. Guaranteeing exactly-once semantics is, in principle, impossible, because you cannot know in general whether the server crashed before or after executing the requested operation.4. Q: 1) Explain how two-phase commit works. (5pt) (chap8)Make sure that you explain (1) coordinator sends vote-request; (2) participants respond; (3) coordinator sends decision; (4) participants ack.2) Explain what happens when a participant, who is in the READY state, times out because it hasn’t received a response from the coordinator yet. (5pt) (chap8)In that case, P can check whether any of the other participants has made a transition to either ABORT or INIT (in which case P can abort) or COMMIT (and commit as well). The difficulty is when all others are in READY: they all need to wait until the coordinator recovers.。