Increasing the Resilience of Distributed and Replicated Database Systems

Abstract

This paper presents a new atomic commitment protocol, enhanced three phase commit (E3PC), that always allows a quorum in the system to make progress. Previously suggested quorum-based protocols (e.g., the quorum-based three phase commit (3PC) (Skeen, 1982)), allow a quorum to make progress in case of one failure. If failures cascade, however, and the quorum in the system is``lost'' (i.e., at a given time no quorum component exists), a quorum can later become connected and still remain blocked. With our protocol, a connected quorum never blocks. E3PC is based on the quorum-based 3PC (Skeen, 1982), and it does not require more time or communication than 3PC. We describe how this protocol can be exploited in a replicated database setting, making the database always available to a majority of the sites. ] 1998 Academic Press

Evaluating the running time of a communication round over the internet

“…The primitive can be initiated by any one of the hosts, called the initiator, and it terminates once information from every host has propagated to all of the hosts. Communication rounds of this sort are employed by many different algorithms and systems, e.g., Byzantine agreement [16], atomic commit [8,23,11], state-machine replication [15], group membership [13], and updates of routing tables. Thus, our study has broad applicability.…”

“…The leader aggregates the information from all the hosts, and sends a message summarizing all the inputs to all the hosts. This algorithm is structured like two-phase commit [8], and like the first two of three communication phases in three-phase commit algorithms, e.g., [23,11]. The algorithm flow is depicted in Figure l(b).…”

Selective data replication for distributed geographical data sets

“…Although there are a few well established patterns and mechanisms available for solving this type of problem, alternatives or derivatives have been constantly studied by many researchers (Schiper & Raynal 1996, Alonso 1997, Keidar & Dolev 1998, JimenezParis et al 2001) to try to further improve performance (Holliday et al 1999, Wiesmann & Schiper 2005) and reduce deadlocks (Holliday et al 2000).…”

Comparison of database replication techniques based on total order broadcast

“…To address this problem, replicated databases based on group communication have been proposed for some time [2], [3], [4]. Those techniques, which typically replace the two-phase commit protocol [5], [6], [7], [8], promise better performance [9], [10], [11], [12], [13], [14], [15], [16], [17], [18] and less deadlocks [19] than classical eager replication, while maintaining strong consistency.…”