S. Sarin and L. Lynch. Discarding obsolete information in a replicated database system. IEEE Transactions on Software Engg., SE(13.1):39--46, Jan. 1987.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....While vector clocks provide a mechanism to represent causality information, the vector clock representation is not readily usable in optimistic protocols due to the forward and backward motion of time. In matrix clock representation, a process maintains a n n matrix of non negative integers [18,21]. A process maintains this matrix as its clock value. This representation has all the properties of vector clocks. In addition, a process P i knows the minimum time value of process P k that is known to every other process P j . This allows processes to discard obsolete information received from ....

S. Sarin and L. Lynch. Discarding obsolete information in a replicated database system. IEEE Transactions on Software Engg., SE(13.1):39--46, Jan. 1987.

....log entries) The actual size of hash is fixed (e.g. 160 bits for SHA1) while the site id could grow depending on the site creation pattern. 3.3. Truncating the Hash History Classical techniques for truncating logs can be applied toward pruning hash histories. The global cutoff timestamp (e.g. [19]) and the acknowledgment timestamp (e.g. 10] can efficiently determine the committed versions; however, these methods fundamentally require one to track the committed state per each site, and hence would not scale to thousands of sites. Instead, we use a simple aging method based on roughly ....

S. K. Sarin and N. A. Lynch. Discarding obsolete information in a replicated database system. IEEE Transactions on Software Engineering, 13(1):39--47, 1987.

....log entries) The actual size of hash is fixed (e.g. 160 bits for SHA1) while the site id could grow depending on the site creation pattern. 3.3. Truncating the Hash History Classical techniques for truncating logs can be applied toward pruning hash histories. The global cutoff timestamp (e.g. [19]) and the acknowledgment timestamp (e.g. 10] can efficiently determine the committed versions; however, these methods fundamentally require one to track the committed state per each site, and hence would not scale to thousands of sites. Instead, we use a simple aging method based on roughly ....

S. K. Sarin and N. A. Lynch. Discarding obsolete information in a replicated database system. IEEE Transactions on Software Engineering, 13(1):39--47, 1987.

....that grows indefinitely as updates accumulate over time. Several techniques are proposed to trim the size of the update log. The first technique deletes all the updates older than a predefined period (e.g. one month) or uses a fixed size update log and rolls over old updates when the log fills up [42, 70]. The second technique is based on an observation that the log trimming problem and the total update ordering problem can both be solved by discovering whether an update is received by all the replicas. Therefore, some systems use a variant of the ack vector algorithm (Section 2.1.2.2) to trim the ....

S. K. Sarin and N. A. Lynch. Discarding obsolete information in a replicated database system. IEEE Transactions on Software Engineering, 13(1):39--47, 1987.

....nodes, designated as the primary node. Network partitions may cause different nodes to act as the primary at 94 different times. Immediate operations are executed with the same set of preceding operations at each node, using a three phase protocol involving all nodes. Bayou session guarantees [Ter94] build on the notion of causal operations, associating updates with sessions so that dependencies can be computed automatically in accordance with user controlled policies; and Bayou s total order for write operations [Pet97] achieves the effect of forced updates. However, immediate updates are ....

Sarin, Sunil K., and Lynch, Nancy A. Discarding obsolete information in a replicated database system. IEEE Transactions on Software Engineering SE-13, No. 1 (January 1987), pp. 39-47

.... Segments Logical matrix clocks [Raynal and Singhal 1996] are traditionally used to discard information in a distributed environment: the componentwise minimum of the columns yields the maximum number of segments per thread that can be discarded (right part of Figure 3) Wuu and Bernstein 1984; Sarin and Lynch 1987]. However, in practice we can discard more information than is indicated by logical matrix clocks as logical clocks capture causality, which is one of the weakest forms of event ordering. In a particular execution, all events are executed is some order (not speci ed by the program) even if they ....

Sarin, S. and Lynch, L. 1987. Discarding obsolete information in a replicated data base system. In IEEE Transactions On Software Engineering, Volume SE, pp. 39-46.

....(meaning they are strongly consistent) a b , C(a) C(b) This means that given two vector clocks C(a) and C(b) it is possible to deduce the causal relation between the two operations a and b. It is possible to augment the dimension of the logical clock even more, resulting in matrix clocks [Wuu84, Sari87]. Matrix clocks provide second order information to a process. It is a list of vector clocks, namely per process the last vector clock that was communicated to the current process. This information can be used to discard obsolete information in distributed systems. 5 Execution replay methods for ....

S. Sarin and L. Lynch. Discarding Obsolete Information in a Replicated Data Base System. In IEEE Transactions on Software Engineering, volume SE, 39--46. January 1987.

....principle is not violated. However, its potential for speedup is limited to situations where the number of events to be processed by each and every process per time step is high; systems whereby timestamps have the form of vectors (e.g. Fidg88] Fidg91] Schm88] Matt88] or matrices (e.g. [Sari87]) Rayn95] CHAPTER 3. MODELLING AND SIMULATION 66 otherwise starvation phenomena will occur as processes remain idle waiting for the clock to advance to the next time step. This approach is more appropriate for shared memory machines, as global synchronization is at present neither easily ....

Sarin, S. K., Lynch, L., "Discarding Obsolete Information in a Replicated Data Base System", IEEE Transactions on Software Engineering, 13, 1 January 1987, pp. 39-46.

....log of updates that grows inde nitely as updates accumulate. Several techniques are proposed to trim the size of the update log. The rst technique deletes all the updates older than a prede ned period (e.g. one month) or uses a xed size update log and rolls over old updates when the log lls up [40, 67]. The second technique is based on an observation that the log trimming problem and the total update ordering problem can both be solved by discovering whether an update is received by all the replicas. Therefore, some systems reuse the ack vector algorithm (Section 2.1.2.2) to trim the update log ....

S. K. Sarin and N. A. Lynch. Discarding obsolete information in a replicated database system. IEEE Transactions on Software Engineering, 13(1):39-47, 1987.

....virtual time scale in accordance with Lamport s clock conditions ( Lamp78] Virtual time is a global temporal coordinate system imposed on a distributed computation. Jefferson visualised virtual time as a one dimensional quantity, but later researchers have proposed vector ( Matt89] and matrix ([Sarin87]) times. A crucial difference between Lamportian and virtual time systems is in the assignment of times to events. In the former, events are assigned times as they occur, whereas in the latter, the times assigned to events are pre determined and fixed. 3.1 Isotach Systems Isotach time is an ....

Sarin, S. K., Lynch, L., Discarding obsolete information in a replicated database system, IEEE Transactions on Software Engineering, Volume 13, Number 1, January 1987.

....if x is an event occurring on site S i , then component [j; k] of ffi mat (x) represents S i s view of S j s view of the progress of S k s local time. The matrix clock satisfies the (S CLK) condition. Matrix clocks have been introduced by Wuu and Bernstein [WB84] and Sarin and Lynch [SL87] as a means to discard obsolete information: since matrix clocks gives information about the other sites views, it makes it possible for a site to stop diffusing an information as soon as it knows that it is in all other sites views. Let us recall the algorithm for computing the matrix clock on ....

....maintain the AG than to directly maintain the n Theta n dimensional matrix clock. This is possible because (1) the AG can be maintained incrementally as described in [EZ92, EZ93] and because (2) it is not necessary to store the whole AG: a garbage collecting algorithm along the lines of [WB84] or [SL87] can be used to discard obsolete events from the AG. 4.1 The algorithm We now describe the incremental algorithm for computing matrix clock on the fly. We assume that every event is tagged with (1) the identifier of the site to which it belongs, and (2) its sequence number on that site. We let ....

[Article contains additional citation context not shown here]

S. K. Sarin and L. Lynch. Discarding obsolete information in a replicated database system. IEEE Trans. on Soft. Eng., SE 13(1):39--46, Jan. 1987.

.... et al. 10] have further proposed a formal approach to recovery by compensating transactions, where they utilize an alternative correctness criteria called predicate wise serializability (PWSR) from [11] The SHARD (System for Highly Available Replicated Data) approach, described in [15] 14] [17] and an associated protocol called the log transformations protocol [3] are designed to support continued database operations in the event of site failures or network partitioning. To achieve this objective, SHARD sacrifices transaction serializability and uses timestamp ordering to ensure ....

S. K. Sarin and N. A. Lynch. Discarding obsolete information in a replicated database system. IEEE Transactions on Software Engineering, SE-13(1):39--47, Jan 1987.

.... algorithms can be used to compute monotonic functions of the global state such as lower bounds on the simulation time (the so called Global Virtual Time or GVT) to which a distributed simulation system has advanced [13] Other applications are checkpointing and recovery of distributed data bases [14, 24] and monitoring and debugging of distributed systems. A snapshot algorithm for systems with FIFO channels was first given by Chandy and Lamport in 1985 [5] The main idea is that immediately after recording the local state, a process sends control messages along each of its (outgoing) channels. ....

Sarin, S., and Lynch, N. Discarding Obsolete Information in a Replicated Database System. IEEE Transactions on Software Engineering SE-13, 1 (1987), 39-47.

....by exploiting semantic properties of transactions. By subdividing transactions into various categories, and utilizing a commutativity property, they demonstrate that communication costs are minimized. The SHARD (System for Highly Available Replicated Data) approach, described in [14] 13] [15] and an associated protocol called the log transformations protocol [2] are designed to support continued database operations in the event of site failures or network partitioning. To achieve this objective, SHARD sacrifices transaction serializability and uses timestamp ordering to ensure ....

S. K. Sarin and N. A. Lynch. Discarding obsolete information in a replicated database system. IEEE Transactions on Software Engineering, SE13 (1):39--47, Jan 1987.

....the matrix timestamp requires O(N 2 ) storage. Reliable multicast schemes [3, 14] use matrix timestamps to ensure reliable delivery in spite of failures, and attach vector timestamps to each multicast message to advance matrix timestamps at other processors. Reliable update propagation schemes [15, 13, 1, 9] must propagate their matrix timestamps in order to advance the matrix timestamps at other processors. If the number of replicas is very large, then the storage and communications overhead of matrix timestamps becomes prohibitive. For example, if there are 10,000 replicas, then the matrix ....

S.K. Sarin and N. Lynch. Discarding obsolete information in a replicated database. IEEE Trans. on Software Engineering, 13(1):39--46, 1987.

No context found.

S.K. Sarin and Nancy A. Lynch. Discarding Obsolete Information in a Replicated Database System. IEEE Transactions on Software Engineering 13:1 (1987), 3947.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC