A Task Migration Mechanism for MPI Applications
نویسندگان
چکیده
Recently, the Cluster of Computers (COC) has been used to run large parallel programs increasingly. Task migration is a desirable and useful facility to implement Load-Balance and High-Availibility in COCs. This paper presents a quick migration protocol for MPI tasks, which allows nonmigrating tasks to execute during most of the time of migration. Process table updating and synchronization are key mechanisms of this protocol. Because MPI does not make provisions for tasks migration, this paper also describes the work required to modify an MPI implementation to allow task migration. At last we introduce our task migration system which is completed grounded on
منابع مشابه
Proactive Fault Tolerance in MPI Applications Via Task Migration
Failures are likely to be more frequent in systems with thousands of processors. Therefore, schemes for dealing with faults become increasingly important. In this paper, we present a fault tolerance solution for parallel applications that proactively migrates execution from processors where failure is imminent. Our approach assumes that some failures are predictable, and leverages the features ...
متن کاملG-javampi: a Grid Middleware for Transparent Mpi Task Migration∗
Resources in a grid are dynamic, heterogeneous, and widely distributed. End users need a simple and efficient way to aggregate and utilize these diverse resources. We introduce a grid middleware called G-JavaMPI, which combines a high-level message passing interface with the Java language to support portable messaging-passing programming in a grid. Different from traditional MPI implementations...
متن کاملCooperative VM Migration: A Symbiotic Virtualization Mechanism by Leveraging the Guest OS Knowledge
A virtual machine (VM) migration is useful for improving flexibility and maintainability in cloud computing environments. However, VM monitor (VMM)-bypass I/O technologies, including PCI passthrough and SR-IOV, in which the overhead of I/O virtualization can be significantly reduced, make VM migration impossible. This paper proposes a novel and practical mechanism, called Symbiotic Virtualizati...
متن کاملThe Hector Distributed Run-Time Environment
The Hector Distributed Run–Time Environment provides a fully integrated run–time environment and scheduling system for MPI programs over networked computer resources. This paper describes the modifications needed to an MPI implementation to make task migration and checkpointing possible, and recent experiments in improved scheduling and optimization. It concludes with a discussion of future wor...
متن کاملVisualizing, measuring, and tuning Adaptive MPI parameters
Adaptive MPI (AMPI) is an advanced MPI runtime environment that offers several features over traditional MPI runtimes, which can lead to a better utilization of the underlying hardware platform and therefore higher performance. These features are overdecomposition through virtualization, and load balancing via rank migration. Choosing which of these features to use, and finding the optimal para...
متن کامل