3Readings Van Steen and Tanenbaum: 5.2 Coulouris: 10.4 L. Lamport, “Time, Clocks and the Ordering of Events in Distributed Systems,” Communications of the ACM, Vol. 21, No. 7, July 1978, ppC.J. Fidge, “Timestamps in Message-Passing Systems that Preserve the Partial Ordering”, Proceedings of the 11th Australian Computer Science Conference, Brisbane, pp , February 1988.
4Problems with Lamport Clocks Lamport timestamps do not capture causality.With Lamport’s clocks, one cannot directly compare the timestamps of two events to determine their precedence relationship.If C(a) < C(b) is not true then a b is also not true.Knowing that C(a) < C(b) is true does not allow us to conclude that a b is true.Example: In the first timing diagram, C(e) = 1 and C(b) = 2; thus C(e) < C(b) but it is not the case that e b
5Problems with Lamport Clocks Why is this important?For the banking example, it doesn’t matter what order operations are executed at each replica.It only matters that all replicas execute the operations in the same order. This is total order.Causal order is used when a message received by a process can potentially affect any subsequent message sent by that process. Those messages should be received in that order at all processes. Unrelated messages may be delivered in any order.Do Lamport timestamps support causal order?NO
6Example Application:Bulletin Board The Internet’s electronic bulletin board (or Google’s Groups or any chat service)Users (processes) join specific groups (discussion groups).Postings, whether they are articles or reactions, are multicast to all group members.Could use a totally-ordered multicasting scheme.Should we be doing this?
7Display from a Bulletin Board Program Users run bulletin board applications which multicast messagesOne multicast group per topic (e.g. os.interesting)Require reliable multicast - so that all members receive messages (more on this later)Ordering:total (makes the numbers the same at all sites)Bulletin board:os.interestingcausal (makes replies come after original message)ItemFromSubject23A.HanlonMachFIFO (gives sender order)24G.JosephMicrokernels25A.HanlonRe: Microkernels26T.L’HeureuxRPC performanceFigure 11.13Colouris27M.WalkerRe: Machend•
8Example Application: Bulletin Board A totally-ordered multicasting scheme does not imply that if message B is delivered after message A, that B is a reaction to A.The receipt of an article precedes the posting of a reaction.The receipt of the reaction to an article should always follow the receipt of the article.
9Example Application: Bulletin Board If we look at the bulletin board example, it is allowed to have items 26 and 27 in different order at different sites.Items 25 and 26 may be in different order at different sites.
10Problem with Lamport Clocks The main problem is that a simple integer clock cannot order both events within a process and events in different processes.C. Fidge developed an algorithm that overcomes this problem.Fidge’s clock is represented as a vector [v1,v2,…,vn] with an integer clock value for each process (vi contains the clock value of process i). This is a vector timestamp.
11Vector Timestamps Properties of vector timestamps vi [i] is the number of events that have occurred so far at PiIf vi [j] = k then Pi knows that k events have occurred at Pj
12Vector Timestamps A vector clock is maintained as follows: Initially all clock values are set to the smallest value (e.g., 0).The local clock value is incremented at least once before each event of interest (in our case this will be a send event) in a process, q i.e., vq[q] = vq[q] +1Let vq be piggybacked on the message sent by process q to process p; We then have:For i = 1 to n dovp[i] = max(vp[i], vq [i] );
13Vector Timestamp For two vector timestamps, va and vb va vb if there exists an i such that va[i] vb[i]va ≤ vb if for all i va[i] ≤ vb[i]va < vb if for all i va[i] ≤ vb[i] AND va is not equal to vbEvents a and b are causally related if va < vb or vb< va .Vector timestamps can be used to guarantee causal message delivery.
14Example Application: Bulletin Board Each process Pi has an array Vi where Vi[j] denotes the number of events that process Pi knows have taken place for Pj.In this application “events” refers to the sending of a message.Thus if Vi[j] = 6 then Pi knows that Pj has sent 6 messages.
15Example Application: Bulletin Board Let Vi be piggybacked on the message sent by process Pi to process Pj ; When process Pj receives the message, then Pj does the following:For k = 1 to n doVj[k] = max(Vi[k], Vj [k] );Thus if process i knows that process k sent 5 messages (Vi [k] =5 and Vj[k] = 3) then process j didn’t know about the latest two messages sent by process k.
16Example Application: Bulletin Board When process i sends a message, it does the following:Vi[i] = Vi[i] + 1;This is basically saying that the number of messages process i has sent is incremented by one.
17Example Application: Bulletin Board When a process Pi posts (sends) an article, a, it multicasts that article with timestamp ViProcess Pj posts a reaction. Let’s call this message r. Assume that the value of the timestamp is VjNote that Vj > ViMessage r may arrive at Pk before message a.Pk will postpone delivery of r to the display of the bulletin board until all messages that causally precede r have been received .
18Example Application: Bulletin Board Message r (from Pj ) is delivered to Pk iff the following conditions are met:Vj[j] = Vk[j]+1This condition is satisfied if r is the next message that Pk was expecting from process PjAssume that Vk[j]=5. This means that Pk knows Pj sent 5 messages and that it should be waiting for the 6th message.If Vj[j] = 8 then messages 6,7 are lost or not yet arrived.Vj[i] ≤ Vk[i] for all i not equal to jThis condition is satisfied if Pk has seen at least as many messages as seen by Pj when it sent message r.Assume for some i that Vj[i] = 2 and Vk[i] =1 and thus Vj[i] > Vk[i] for some i. This indicates that Pi sent a message that was received by Pj but not Pk
19Example Application: Bulletin Board [0,0,0]P1[0,0,0]P3[0,0,0]Post a[1,0,0]a[1,0,0]e[1,0,0]c[1,0,1]r: Reply adgb[1,0,1]Note that if two events happen without any message, then we can't say anything about their relative occurrence in time.In this example a <= b <= c <= d <= f, but we can say little about e other than e <= f. We say that events such as a and e are concurrent.Note also that “happened before” does not demonstrate causality. There is merely the potential for causality.[1,0,1]Message a arrives at P2 before the reply r from P3 does
20Example Application: Bulletin Board [0,0,0]P1[0,0,0]P3[0,0,0]Post a[1,0,0]ad[1,0,0]g[1,0,1]r: Reply aBufferedbNote that if two events happen without any message, then we can't say anything about their relative occurrence in time.In this example a <= b <= c <= d <= f, but we can say little about e other than e <= f. We say that events such as a and e are concurrent.Note also that “happened before” does not demonstrate causality. There is merely the potential for causality.c[1,0,1][1,0,0]Deliver rThe message a arrives at P2 after the reply from P3; The reply isnot delivered right away.
21Summary No notion of a globally shared clock. Local (physical) clocks must be synchronized based on algorithms that take into account network latency.Knowing the absolute time is not necessary.Logical clocks can be used for ordering purposes.