Scalable resource management in high performance computers.
- Eitan
- Fabrizio
- Juan
- Salvador
Clusters of workstations have emerged as an important platform for building cost-effective, scalable and highly-available computers. Although many hardware solutions are available today, the largest challenge in making large-scale clusters usable lies in the system software. In this paper we present STORM, a resource management tool designed to provide scalability, low overhead and the flexibility necessary to efficiently support and analyze a wide range of job scheduling algorithms. STORM achieves these feats by closely integrating the management daemons with the low-level features that are common in state-of-the-art high-performance system area networks. The architecture of STORM is based on three main technical innovations. First, a sizable part of the scheduler runs in the thread processor located on the network interface. Second, we use hardware collectives that are highly scalable both for implementing control heartbeats and to distribute the binary of a parallel job in near-constant time, irrespective of job and machine sizes. Third, we use an I/O bypass protocol that allows fast data movements from the file system to the communication buffers in the network interface and vice versa. The experimental results show that STORM can launch a job with a binary of 12MB on a 64 processor/32 node cluster in less than 0.25 sec on an empty network, in less than 0.45 sec when all the processors are busy computing other jobs, and in less than 0.65 sec when the network is flooded with a background traffic. This paper provides experimental and analytical evidence that these results scale to a much larger number of nodes. To the best of our knowledge, STORM is at least two orders of magnitude faster than existing production schedulers in launching jobs, performing resource management tasks and gang scheduling.
- Research Organization:
- Los Alamos National Laboratory (LANL), Los Alamos, NM (United States)
- Sponsoring Organization:
- USDOE
- OSTI ID:
- 976130
- Report Number(s):
- LA-UR-02-1672; TRN: US201009%%561
- Resource Relation:
- Conference: Submitted to: International Supercomputing Conference, New York, June, 2002
- Country of Publication:
- United States
- Language:
- English
Similar Records
Adaptive Parallel Job Scheduling with Flexible CoScheduling
A New coscheduling technique for a cluster of symmetric multiprocessors