A Deterministic Database Replication Protocol Where Multicast Writesets Never Get Aborted
Several approaches for the full replication of data in distributed databases [1] have been studied. One of the preferred techniques is the eager update everywhere based on the total-order multicast delivery service [2], where the most outstanding varietie
- PDF / 127,083 Bytes
- 2 Pages / 430 x 660 pts Page_size
- 112 Downloads / 211 Views
ad P´ublica de Navarra, 31006 Pamplona, Spain {jr.juarez,enrique.armendariz,mendivil,joserra}@unavarra.es 2 Instituto Tecnol´ogico de Inform´atica, 46022 Valencia, Spain [email protected]
Introduction. Several approaches for the full replication of data in distributed databases [1] have been studied. One of the preferred techniques is the eager update everywhere based on the total-order multicast delivery service [2], where the most outstanding varieties are: certification-based and weak-voting [1]. Under this approach, the execution flow of a transaction can be split into two different main phases: the first one, all operations are entirely executed at the delegate replica of the transaction; and followed by the second phase, started when the transaction requests its commit, all updates are collected and grouped (denoted as writeset) at the delegate replica and sent to all replicas. The commitment or abortion of a transaction is decided upon the delivery of the message. In the case of certification-based ones, each replica holds an ordered log of already committed transactions and the writeset is certified [3], against the log, to commit or abort the transaction. On the other hand, weak-voting ones atomically apply the delivered writeset at remote replicas whilst the delegate, if it is still active, reliably multicasts [2] a commit message. Thus, the certification-based presents a better behavior in terms of performance, only one message is multicast per transaction, but with higher abortion rates [1]. Recently, due to the use of DBMS providing SI, we have found several certification-based protocols to achieve, actually a weaker form called GSI [3], this isolation level in a replicated setting [3] while quite a few weak-voting ones [4]. Total-order multicast in database replication offers two attractive properties: a) they reliably send the writeset to all replicas; and, b) they provide the same scheduling of transactions and, thus, all replicas reach the same decision for each transaction in the replicated setting. However, if we focus in the scheduling policy, there is a point where certification-based and weak-voting techniques converge: all delivered messages coming from a replica are known to be successfully certified (i.e no need of a log for certification ones) and committed (i.e. no need for the additional message exchange for weak-voting ones). Nevertheless, this may cause the penalization of certain transaction patterns. We propose an eager update-everywhere replication protocol [5] that follows the most straightforward scheduling policy: a round-robin based on replica identifiers that is unique and known by all replicas. At a given slot, only those writesets coming from the replica associated to that slot are allowed to commit, while the conflicting local transactions should be aborted. This generates a unique scheduling configuration of all replicas where all writesets are applied in the same order providing GSI.
This work has been partially supported by the EU FEDER and Spanish MEC under grant TIN2
Data Loading...
