Conference Paper

Improving Recovery in Weak-Voting Data Replication.

DOI: 10.1007/978-3-540-76837-1_17 Conference: Advanced Parallel Processing Technologies, 7th International Symposium, APPT 2007, Guangzhou, China, November 22-23, 2007, Proceedings
Source: DBLP


Nowadays eager update everywhere replication protocols are widely proposed for replicated databases. They work together with
recovery protocols in order to provide highly available and fault-tolerant information systems. This paper proposes two enhancements
for reducing the recovery times, minimizing the recovery information to transfer. The idea is to consider on one hand a more
realistic failure model scenario –crash recovery with partial amnesia– and on the other hand to apply a compacting technique.
Moreover, it is provided amnesia support avoiding possible state inconsistencies –associated to the failure model assumed–
before starting the recovery process at recovering replicas.

Download full-text


Available from: José Enrique Armendáriz-Iñigo,
  • Source
    • "CORE A 7 [13] [19] [24] [41] [43] [44] [50] CORE B 19 [1] [2] [3] [4] [6] [7] [11] [15] [18] [21] [22] [25] [27] [31] [32] [34] [38] [40] [51] LNCS 9 [9] [12] [16] [28] [33] [37] [47] [49] [53] Other 17 [10] [14] [17] [20] [23] [26] [29] [30] [35] [36] [42] [45] [46] [48] [52] Table 2: Project publications an approach that allows to reorganise the transaction completion order for reducing the probability of abortion due to constraint violations detected by integrity checking. To this end, we have proposed prioritised atomic multicast protocols [40] [41], and have designed a middleware module that can rate transactions according to their likelihood of integrity violations. "
    [Show abstract] [Hide abstract]
    ABSTRACT: Nowadays, there are good research results in the fields of dependability and data consistency in distributed static environments, as proven by recent publications on secure and replicated databases. However, even in those environments there exist some issues that can be improved, such as their recovery protocols and the design of frameworks for maintaining concurrently multiple replication protocols specially tailored for the current set of applications that access a given replicated database, or for exchanging at runtime those protocols. Besides this, new and improved mobile devices exist, and wireless networks may be found and used anywhere, so their users will ask soon for dependable collaborative applications. Thus, the distributed protocols run by these mobile devices will be one of the key issues for improving their dependability, being distributed agreement the most important. This project is focused on: (1) the improvements described above for static environments; (2) the design, specification and implementation of a dependable architecture for dynamic environments; (3) new supports for group membership, distributed agreement, secure communications, authentication, and consistency and replication management in those dynamic environments.
  • Source
    • "The author has participated in the following papers of this project [10] [25] [26] [27] [29] [30] [32] [41]. These papers will be explained in the following section. "