When Scalability Meets Consistency: Genuine Multiversion Update-Serializable Partial Data Replication

Abstract

Abstract-In this article we introduce GMU, a genuine partial replication protocol for transactional systems, which exploits an innovative, highly scalable, distributed multiversioning scheme. Unlike existing multiversion-based solutions, GMU does not rely on a global logical clock, which represents a contention point and can limit system scalability. Also, GMU never aborts read-only transactions and spares them from distributed validation schemes. This makes GMU particularly efficient in presence of read-intensive workloads, as typical of a wide range of real-world applications.GMU guarantees the Extended Update Serializability (EUS) isolation level. This consistency criterion is particularly attractive as it is sufficiently strong to ensure correctness even for very demanding applications (such as TPC-C), but is also weak enough to allow efficient and scalable implementations, such as GMU. Further, unlike several relaxed consistency models proposed in literature, EUS has simple and intuitive semantics, thus being an attractive, scalable consistency model for ordinary programmers.We integrated the GMU protocol in a popular open source in-memory transactional data grid, namely Infinispan. On the basis of a large scale experimental study performed on heterogeneous experimental platforms and using industry standard benchmarks (namely TPC-C and YCSB), we show that GMU achieves linear scalability and that it introduces negligible overheads (less than 10%), with respect to solutions ensuring non-serializable semantics, in a wide range of workloads.

Scalable service-oriented replication with flexible consistency guarantee in the cloud

“…However, unlike our proposal, their approach is driven from the eventual consistency model, where the variation is that the solution is able to change the extent to which consistency can be violated. Instead of providing different consistency models to various scenarios, there also exist a consistency model that is static, but can cope with most scenarios of consistency and scalability requirement [34]. Such model is extended from the eventual consistency model, it is designed to be sufficiently strong to ensure most scenarios that demand good consistency, but also weak enough to allow good scalability.…”

Self-Tuning Transactional Data Grids: The Cloud-TM Approach

“…In order to maximize performance, Infinispan maintains data in-memory, and achieves fault-tolerance via data replications rather than via disk logging mechanisms. To achieve strong consistency without sacrificing scalability, Infinispan integrates GMU [7], a novel, fully decentralized, multi-versioning algorithm that achieves high scalability by means of genuine replication techniques (which ensure that only the nodes replicating data accessed by the transaction are involved in its processing), and by avoiding to ever abort or block read-only transactions.…”

“…Cloud-TM, conversely, ensures strong consistency by leveraging on the abstraction of transaction [6]. Transactional consistency and scalability, two properties often seen as antagonists, are reconciled in the Cloud-TM platform thanks to innovative transactional consistency schemes [7], [8] designed precisely to meet the scalability and elasticity requirements of typical cloud infrastructures.…”

Performance Modelling of Partially Replicated In-Memory Transactional Stores

“…As for consistency, first generation of NoSQL stores [3] embraced very weak consistency models, like eventual consistency. However, the inherent complexity of building applications on top of weakly consistent systems has been recently recognized by some of the pioneers of eventual consistency [4], and motivated several works providing stronger consistency guarantees [2,5]- [7].…”