Cluster scheduling for explicitly-speculative tasks

Abstract

Public reporting burden for the collection of information is estimated to average 1 hour per response, including the time for reviewing instructions, searching existing data sources, gathering and maintaining the data needed, and completing and reviewing the collection of information. Send comments regarding this burden estimate or any other aspect of this collection of information, including suggestions for reducing this burden, to Washington Headquarters Services, Directorate for Information Operations and Reports, 1215 Jefferson Davis Highway, Suite 1204, Arlington VA 22202-4302. Respondents should be aware that notwithstanding any other provision of law, no person shall be subject to a penalty for failing to comply with a collection of information if it does not display a currently valid OMB control number. AbstractA process scheduler on a shared cluster, grid, or supercomputer that is informed which submitted tasks are possibly unneeded speculative tasks can use this knowledge to better support increasingly prevalent user work habits, lowering user-visible response time, lowering user costs, and increasing resource provider revenue.Large-scale computing often consists of many speculative tasks (tasks that may be canceled) to test hypotheses, search for insights, and review potentially finished products. For example, speculative tasks are issued by bioinformaticists comparing dna sequences, computer graphics artists rendering scenes, and computer researchers studying caching. This behaviorexploratory searches and parameter studies, made more common by the costeffectiveness of cluster computing -on existing schedulers without speculative task support results in a mismatch of goals and suboptimal scheduling. Users wish to reduce their time waiting for needed task output and the amount they will be charged for unneeded speculation, making it unclear to the user how many speculative tasks they should submit. This thesis introduces 'batchactive' scheduling (combining batch and interactive characteristics) to exploit the inherent speculation in common application scenarios. With a batchactive scheduler, users submit explicitlylabeled batches of speculative tasks exploring ambitious lines of inquiry, and users interactively request task outputs when these outputs are found to be needed. After receiving and considering an output for some time, a user decides whether to request more outputs, cancel tasks, or disclose new speculative tasks. Users are encouraged to disclose more computation because batchactive scheduling intelligently prioritizes among speculative and non-speculative tasks, providing good wait-time-based metrics, and because batchactive scheduling employs an incentive pricing mechanism which charges for only requested task outputs (i.e., unneeded speculative tasks are not charged), providing better cost-based metrics for users. These aspects can lead to higher billed server utilization, encouraging batchactive adoption by resource providers organized as either cost-or profit-centers. vi · Cluster sche...

Service contracts and aggregate utility functions

“…This is a well-trodden field: the Alpha OS [8] handled time-varying utility functions; [6] discussed how to do processor scheduling for them; [18] looked at tradeoffs between multiple applications with multiple utility-dimensions; Muse [5] and Unity [24] used utility functions for continuous fine grained service quality control; and [20] described using utility functions to allow speculative execution of tasks. Unlike prior systems, our clients use aggregate utility functions to control service provider behavior across multiple jobs.…”

Two Auction‐Based Resource Allocation Environments: Design and Experience

“…The idea of scheduling based on per-job utility functions is a well-trodden field: the Alpha OS [19] handles time-varying utility functions; Chen and Muhlethaler [14] discuss how to do processor scheduling for them; Lee et al [31] look at trade-offs between multiple applications with multiple utility-dimensions; Muse [13] and Unity [48] use utility functions for continuous fine grained service quality control; and Petrou et al [40] describe using utility functions to allow speculative execution of tasks. Siena [10] considers the impact of user aggregate utility functions to control service provider behavior across multiple jobs.…”

A (condensed) parametric study of optimistic computation in wide-area, distributed environments

“…At the hardware level, speculative branch execution is used to prevent bubbles from getting into an instruction pipeline. At the system level, optimism can be used to do speculative cluster scheduling to service applications such as speculative DNA sequencing [13]. At the computational domain level, optimistic simulation has been long known to produce performance benefits [10,9], under the right conditions, and is still finding applications today [18].…”