A. Wool. Quorum systems in replicated databases: science or fiction. Bull. IEEE Technical Committee on Data Engineering, 21(4):3--11. IEEE, December 1998.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....of majority voting Majority voting has been proposed as a simple yet robust replication method for quite a while [6, 12] but no system has used it in a high throughput environment. The oftencited reason is that it is inefficient, because read requests must contact multiple remote replicas [14]. This reason, however, does not apply to FAB for the two reasons. First, we apply an optimistic read technique for the common case scenario of reading from a (logical) block that is already consistent. Here, the coordinator reads the actual block contents from one idle replica and reads only ....

Avishai Wool. Quorum systems in replicated databases: science or fiction? Bull. IEEE Technical Committee on Data Engineering, 21(4), December 1998.

....For example, in the grid protocol [3] the quorum size is approximately (e.g. in a system with 100 nodes the quorum size is 20) Therefore, it is possible to provide very high availability with a reasonable communication cost. Additionally, quorum systems can be used to execute loadbalancing [18]: as only a small fraction of servers receive each request, if the quorum selection strategy is properly chosen, each server will have to handle only a small fraction of requests. In [9] the authors have proposed a quorum system that presents asymptotically optimal availability (the availability ....

A. Wool. Quorum systems in replicated databases: Science or fiction?. Bulletin of the IEEE Technical Committee on Data Engineering, December 1998. 9

....of (at least) a majority of servers to maintain mutual exclusion properties. To the best of our knowledge, Deno is the only voting scheme that allows for light weight replica creation and retirement, requiring the participation of only two servers. 16 Finally, we note that recent work [42] investigated why voting systems (i.e. quorums) have yet to become widespread in real world applications. One of the conclusions is that voting does not enhance availability because either failures are positively correlated (when servers are on a single LAN) or network partitions occur ....

A. Wool, Quorum Systems in Replicated Databases: Science or Fiction?, Bulletin of the Technical Committee on Data Engineering, vol. 21, pp. 3-11, 1998.

....the participation of (at least) a majority of servers to maintain mutual exclusion properties. To the best of our knowledge, Deno is the only voting scheme that allows for light weight replica creation and retirement, requiring the participation of only two servers. We note that recent work [25] investigated why voting systems (i.e. quorums) have yet to become widespread in real world applications. One of the conclusions is that voting does not enhance availability because either failures are positively correlated (when servers are on a single LAN) or network partitions occur (when ....

A. Wool, "Quorum Systems in Replicated Databases: Science or Fiction?," Bulletin of the Technical Committee on Data Engineering, vol. 21, pp. 3-11, 1998.

....primary copy model. In conclusion, in spite of common attacks on the practical bene ts of replication, we believe that broadcast support holds the promise of actually delivering these bene ts. Recently, other researchers have started examining novel requirements for replication. For example, Wool [21] argues that quorum based replication deserves a new look as new technological advances and di erent application requirements are brought to light. Similarly, we argue that broadcast based replication should be revisited, now that the performance of broadcast systems has improved. The protocols we ....

A. Wool, \Quorum Systems in Replicated Databases: Science or Fiction?" Bulletin of the IEEE Computer Society Technical Committee on Data Engineering, Dec. 1998. 19

.... different cache invalidation strategies for mobile wireless environments [8] Huang et al. analyzed different data replication methods with the objective of minimizing the communication cost between the mobile and stationary hosts in a mobile environment [21] Finally, we note that recent work [36] investigated why voting systems (i.e. quorums) have yet to become widespread in real world applications. One of the conclusions is that voting does not enhance availability because either failures are positively correlated (when servers are on a single LAN) or network partitions occur (when ....

A. Wool, "Quorum Systems in Replicated Databases: Science or Fiction?," Bulletin of the Technical Committee on Data Engineering, vol. 21, pp. 3-11, 1998.

....all currency to a single server emulates a primary copy scheme. We present results showing the rate at which both schemes commit updates, as well as the performance advantages of using currency proxies to temporarily reallocate currency during planned disconnections. We note that recent work [23] has investigated why quorum systems have yet to become widespread in real world applications. One of the conclusions is that quorums do not enhance availability because either failures are positively correlated (when servers are on a single LAN) or network partitions occur (when servers are ....

A. Wool, "Quorum Systems in Replicated Databases: Science or Fiction?" Bulletin of the Technical Committee on Data Engineering, vol. 21, pp. 3-11, 1998.

....systems (e.g. Argus [22] and Thor [23] and shared memory emulation systems (e.g. Ivy [21] and Munin [5] However, accessing only a single replica precludes any ability to mask arbitrary server failures and therefore is unsuitable for our survivability goals. Moreover, as recently discussed in [41], the communication overhead of accessing multiple servers has been steadily decreasing with improvements in network technology, and thus in some cases these costs may no longer outweigh the benefits of reduced overall server load that quorums can offer. One of the primary focuses of this paper is ....

A. Wool. Quorum systems in replicated databases: Science or fiction? Bull. IEEE Technical Committee on Data Engineering, 21(4):3--11, December 1998.

No context found.

A. Wool. Quorum systems in replicated databases: science or fiction. Bull. IEEE Technical Committee on Data Engineering, 21(4):3--11. IEEE, December 1998.

No context found.

WOOL, A. 1998. Quorum systems in replicated databases: Science or fiction? Data Eng. Bull. 21,4 (Dec.), 3--11.

No context found.

A. Wool. Quorum systems in replicated databases: science or fiction. Bull. IEEE Technical Committee on Data Engineering, 21(4):3--11. IEEE, December 1998.

No context found.

Avishai Wool. Quorum systems in replicated databases: science or fiction? Bull. IEEE Technical Committee on Data Engineering, 21(4):3--11, December 1998.

No context found.

A. Wool. Quorum systems in replicated databases: Science or fiction? Bulletin of the Techn. Committee on Data Engineering, 21(4), 1998.

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