BTMD: Small, Fast Diffs for WAN-Based DSM
نویسندگان
چکیده
The performance of distributed shared memory (DSM) over a wide area network (WAN) may suffer due to high network latencies because of the low bandwidth available. To achieve acceptable performance, WAN-based DSM must minimize network traffic. DSM systems transmit consistency information over the network via diffs. Existing diff mechanisms for DSM systems that run on local area networks are not appropriate for WAN-based DSM systems because they do not compute minimally sized diffs. This paper presents a new, low-overhead diff mechanism for WAN-based DSM systems. Specifically, the diff mechanism quickly computes diffs, quickly applies diffs, and minimizes the space used to store and transmit diffs. We show that our mechanism exhibits significantly reduced memory consumption and network costs when compared to existing diff mechanisms.
منابع مشابه
The Affinity Entry Consistency Protocol
In this paper we propose a novel software-only distributed sharedmemory system (SW-DSM), the Affinity Entry Consistency (AEC) protocol. The protocol is based on Entry Consistency but, unlike previous approaches, does not require the explicit association of shared data to synchronization variables, uses the page as its coherence unit, and generates the set of modifications (in the form of diffs)...
متن کاملAdapting Sequence Models for Sentence Correction
In a controlled experiment of sequence-tosequence approaches for the task of sentence correction, we find that characterbased models are generally more effective than word-based models and models that encode subword information via convolutions, and that modeling the output data as a series of diffs improves effectiveness over standard approaches. Our strongest sequence-to-sequence model improv...
متن کاملAn Efficient Lock Protocol for Home-Based Lazy Release Consistency
Home-based Lazy Release Consistency (HLRC) shows poor performance on lock based applications because of two reasons: (1) a whole page is fetched on a page fault while actual modification is much smaller, and (2) a home is at the fixed location while access pattern is migratory. In this paper we present an efficient lock protocol for HLRC. In this protocol, the pages that are expected to be used...
متن کاملLaunch and Iterate: Reducing Prediction Churn
Practical applications of machine learning often involve successive training iterations with changes to features and training examples. Ideally, changes in the output of any new model should only be improvements (wins) over the previous iteration, but in practice the predictions may change neutrally for many examples, resulting in extra net-zero wins and losses, referred to as unnecessary churn...
متن کاملp-Branes as Antisymmetric Nonabelian Tensorial Gauge Field Theories of Diffeomorphisms in p + 1 dimensions
Long ago, Bergshoeff, Sezgin, Tanni and Townsend have shown that the light-cone gauge-fixed action of a super p-brane belongs to a new kind of supersymmetric gauge theory of p-volume preserving diffeomorphisms (diffs) associated with the p-spatial dimensions of the extended object. These authors conjectured that this new kind of supersymmetric gauge theory must be related to an infinite-dim non...
متن کامل