Conference Paper

Commit Protocols for a Distributed Transactional Memory

Inst. fur Inf., Heinrich-Heine-Univ. Dusseldorf, Düsseldorf, Germany
DOI: 10.1109/PDCAT.2010.74 Conference: Parallel and Distributed Computing, Applications and Technologies (PDCAT), 2010 International Conference on
Source: IEEE Xplore

ABSTRACT Concurrency control in distributed and parallel applications has been studied for many years but is still an ongoing research topic. Transactional memory addresses this challenge for multicore processors by proposing to execute critical sections as restartable transactions combined with optimistic synchronization. Thus the programmer has not to reason about complex lock management and deadlocks. We believe that some of these ideas are also useful for distributed systems. Therefore, we are developing the Object Sharing Service (OSS) providing transparent data sharing for clusters and grids. OSS supports different consistency models for replica management within one application. In this paper we present the design and implementation of different transaction commit protocols for supporting transactional consistency. The main challenge of the resulting distributed transactional memory (DTM) is how to mask network latency allowing to commit transactions fast. Experiments with synthetic micro benchmarks and a Map Reduce application on the Grid'5000 platform show that a DTM is efficiently providing strong consistency for shared data.

0 Followers
 · 
90 Views
  • [Show abstract] [Hide abstract]
    ABSTRACT: Cloud computing enables diverse new application areas for distributed computing. Many upcoming cloud applications do not fit to simple programming models such as "embarrassing parallelism" but have complex data dependencies and require atomic operations spanning multiple objects. Some large-scale storage systems already implement atomic multiobject operations, but they do not address the complementary problem of efficiently propagating replica updates. In this paper, we present the design and implementation of a smart replication protocol in the ECRAM in-memory storage, which supports atomic multi-object operations. The performance analysis shows that the adaptive mechanism requires much less bandwidth, less memory, and results in improved application performance and responsiveness.
    Parallel and Distributed Systems (ICPADS), 2012 IEEE 18th International Conference on; 01/2012