Revisiting 1-copy equivalence in clustered databases

Abstract

Recently renewed interest in scalable database systems for shared nothing clusters has been supported by replication protocols based on group communication that are aimed at seamlessly extending the native consistency criteria of centralized database management systems. By using a read-one/write-all-available approach and avoiding the fine-grained synchronization associated with traditional distributed locking, one needs just a single distributed interaction step for each update transaction. Therefore the system can easily be scaled to a large number of replicas, especially, with read intensive loads typical of Web server support environments.In this paper we point out that 1-copy equivalence for causal consistency, which is subsumed by both serializability and snapshot isolation criteria, depends on basic session guarantees that are costly to ensure in clusters, especially in a multi-tier environment. We then point out a simple solution that guarantees causal consistency in the Database State Machine protocol and evaluate its performance, thus highlighting the cost of seamlessly providing common consistency criteria of centralized databases in a clustered environment. BACKGROUNDA number of group communication based database replication protocols have been proposed, in particular, considering a readone/write-all-available approach and a single interaction for each update transaction [13,14,19,12]. We focus on protocols that exploit optimistic execution and in particular the Database State Machine (DBSM) [14,7,20]. These protocols are multi-master, transactions can be submitted to and executed by several replicas, and follow the passive replication paradigm [5,9] in which each transaction is executed by one of the replicas and its state changes propagated to the other replicas.A transaction is immediately executed by the replica to which it is submitted without any a priori coordination. Locally, transactions are synchronized according to the specific concurrency control mechanism of the database engine and thus according to the centralized consistency criteria implemented.Read-only transactions are simply executed locally at the database Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. To copy otherwise, to republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee. to which they are submitted and thus controlled strictly by the local concurrency control mechanism. In contrast, update transactions are not readily committed upon receiving the commit request. Instead, the tuples read (read-set) and written (write-set) are gathered and a termination protocol initiated. The goal of the termination protocol is to decide the order and the outcome of the transaction such that the global correctness criteria is satisfied. This is achieve...

A formal characterization of SI-based ROWA replication protocols

“…In the meantime, the transaction associated to such writeset would have been able to commit at some sites. Due to this, a given client is able to perceive a transaction inversion (i.e., new read requests are retrieving older values than past read requests) if some of its read accesses are forwarded to different server replicas [31]. This explains why in our Definition 7, the commit/write time assigned to the commit or write operations in a 1MV-schedule is that of the first replica that has executed such operations.…”

Practical Database Replication

“…Group-based replication protocols take advantage of the specific properties of the group communication primitives, such as ordering and atomicity of messages, to eliminate the possibility of deadlocks, reduce message overhead, and increase performance. These primitives are key components to foster the adoption of software-based replication on shared-nothing architectures with commodity machines, and Chapter 3 describes in detail this family of protocols (e.g., [42,76,84,99,104,111,129]…”

Correctness Criteria for Database Replication: Theoretical and Practical Aspects

“…Strong session serializability prevents transactions reordering within a client's session. Causality expected by the clients is studied in [19] as well. Differently from us, [19] considers only Besides uniform characterization of consistency degrees, we also show how each of them can be achieved in the context of BaseCON and two other replication solutions.…”

“…Causality expected by the clients is studied in [19] as well. Differently from us, [19] considers only Besides uniform characterization of consistency degrees, we also show how each of them can be achieved in the context of BaseCON and two other replication solutions. BaseCON was originally inspired by conflict-aware scheduling, a replication technique by Amza et al [20].…”