Dependency-Aware Rollback and Checkpoint-Restart for Distributed Task-Based Runtimes

نویسندگان

  • Kiril Dichev
  • Herbert Jordan
  • Konstantinos Tovletoglou
  • Thomas Heller
  • Dimitrios S. Nikolopoulos
  • Georgios Karakonstantis
  • Charles Gillan
چکیده

With the increase in compute nodes in large compute platforms, a proportional increase in node failures will follow. Many application-based checkpoint/restart (C/R) techniques have been proposed for MPI applications to target the reduced mean time between failures. However, rollback as part of the recovery remains a dominant cost even in highly optimised MPI applications employing C/R techniques. Continuing execution past a checkpoint (that is, reducing rollback) is possible in message-passing runtimes, but extremely complex to design and implement. Our work focuses on task-based runtimes, where task dependencies are explicit and message passing is implicit. We see an opportunity for reducing rollback for such runtimes: we explore task dependencies in the rollback, which we call dependency-aware rollback. We also design a new C/R technique, which is influenced by recursive decomposition of tasks, and combine it with dependency-aware rollback. We expect the dependency-aware rollback to cancel and recompute less tasks in the presence of node failures. We describe, implement and validate the proposed protocol in a simulator, which confirms these expectations. In addition, we consistently observe faster overall execution time for dependency-aware rollback in the presence of faults, despite the fact that reduced task cancellation does not guarantee reduced overall execution time.

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

منابع مشابه

An Efficient Recovery Mechanism with Checkpointing Approach for Cluster Federation

Checkpoint and recovery protocols are commonly used in distributed applications for providing fault tolerance. A distributed system may require taking checkpoints from time to time to keep it free of arbitrary failures. In case of failure, the system will rollback to checkpoints where global consistency is preserved. Checkpointing is one of the fault-tolerant techniques to restore faults and to...

متن کامل

CRAK: Linux Checkpoint/Restart As a Kernel Module

Process checkpoint/restart is a very useful technology for process migration, load balancing, crash recovery, rollback transaction, job controlling and many other purposes. Although process migration has not yet been widely used and is not widely available commercial systems, the growing shift of computing facilities from supercomputers to networked workstations and distributed systems is incre...

متن کامل

Tight Upper Bound on Useful Distributed System Checkpoints

In this paper, we give an alternative proof of the necessary and sufficient condition for achieving optimal checkpoint garbage collection in distributed systems [9]. We show that, by formulating the recovery line calculation problem as a reachability analysis problem on a rollback-dependency graph, the proof can be greatly simplified. We also present a polynomial-time optimal garbage collection...

متن کامل

More Properties of Communication-Induced Checkpointing Protocols with Rollback-Dependency Trackability

Rollback-Dependency Trackability (RDT) is a property stating that all rollback dependencies between local checkpoints are on-line trackable using a transitive dependency vector. In this paper, we introduce some properties of communication-induced checkpointing protocols possessing the RDT property. First, we demonstrate that wherever an RDT protocol detects a PCM-path in the checkpoint and comm...

متن کامل

Rollback Recovery Scheme for Distributed Shared Memory Clusters

In this paper, an unified lightweight error recovery scheme based on coordinated checkpointing and rollback for distributed shared memory clusters is proposed. The new scheme maintains multiple globally consistent checkpoints of the state of a distributed shared memory cluster and recovers to a pre-fault checkpoint of the system. It also describes and evaluates the coordinated checkpointing. Th...

متن کامل

ذخیره در منابع من


  با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

عنوان ژورنال:
  • CoRR

دوره abs/1705.10208  شماره 

صفحات  -

تاریخ انتشار 2017