D2STM: Dependable Distributed Software Transactional Memory

Abstract

Software Transactional Memory (STM) systems have emerged as a powerful paradigm to develop concurrent applications. At current date, however, the problem of how to build distributed and replicated STMs to enhance both dependability and performance is still largely unexplored. This time by a non-blocking distributed certification scheme, which we name BFC (Bloom Filter Certification). BFC exploits a novel Bloom Filter-based encoding mechanism that permits to significantly reduce the overheads of replica coordination at the cost of a user tunable increase in the probability of transaction abort. Through an extensive experimental study based on standard STM benchmarks we show that the BFC scheme permits to achieve remarkable performance gains even for negligible (e.g. 1%) increases of the transaction abort rate.

AGGRO: Boosting STM Replication via Aggressively Optimistic Transaction Processing

“…Finally, our work is clearly related to the recent literature on distributed STMs [4,6,15,16]. However, except [6], none of these solutions leverages on replication in order to enhance system dependability.…”

“…However, except [6], none of these solutions leverages on replication in order to enhance system dependability. In AGGRO, on the other hand, dependability represents a first class design goal, and the STM performance is optimized by seeking maximum overlap between the non-blocking OAB-based replica coordination phase and local transaction processing activities.…”

“…The traces were collected running two benchmark applications, RB-Tree and List, that were originally used for evaluating DSTM2 [13] and, later on, adopted in a number of performance evaluation studies of STM systems [5,6]. These applications perform repeated insertion, removal and search operations of a randomly chosen integer in a set of integers implemented either as a sorted single-linked list or as a red-black tree.…”

Identifying the Optimal Level of Parallelism in Transactional Memory Applications

“…When considering DTM systems [17], the degree of concurrency of the platform, which we call also global multiprogramming level, is determined not only by the number of threads deployed on each node, but also by the number of nodes composing the platform. We note that in distributed settings, purely explorationbased techniques, like the one described in Section 3, are much less effective for two main reasons.…”

SPECULA: Speculative Replication of Software Transactional Memory

“…As a direct consequence, state of the art transactional replication mechanisms would introduce a significant overhead when employed in TM contexts. This phenomenon is clearly highlighted by the experimental data reported in Figure 1, which ware gathered using D 2 ST M , a state of the art Java-based DTM platform [4] that integrates an Atomic Broadcast (AB) based replication protocol [5], [6]. This DTM platform was evaluated using a standard benchmark for Software TM (STM) systems, namely STMBench7 [7], which was deployed on a cluster of up to 8 nodes 1 .…”