Iso-Level CAFT: How to Tackle the Combination of Communication Overhead Reduction and Fault Tolerance Scheduling

Abstract

To schedule precedence task graphs in a more realistic framework, we introduce an efficient fault tolerant scheduling algorithm that is both contention-aware and capable of supporting ε arbitrary fail-silent (failstop) processor failures. The design of the proposed algorithm which we call Iso-Level CAFT, is motivated by (i) the search for a better loadbalance and (ii) the generation of fewer communications. These goals are achieved by scheduling a chunk of ready tasks simultaneously, which enables for a global view of the potential communications. Our goal is to minimize the total execution time, or latency, while tolerating an arbitrary number of processor failures. Our approach is based on an active replication scheme to mask failures, so that there is no need for detecting and handling such failures. Major achievements include a low complexity, and a drastic reduction of the number of additional communications induced by the replication mechanism. The experimental results fully demonstrate the usefulness of Iso-Level CAFT.