Ananke: A Q-Learning-Based Portfolio Scheduler for Complex Industrial Workflows

Ma, Shenjun; Ilyushkin, Alexey; Stegehuis, Alexander; Iosup, Alexandru

doi:10.1109/icac.2017.21

Cited by 6 publications

(10 citation statements)

References 38 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Private/hybrid clouds such as the DAS [15] are often homogeneous [32] or provide a homogeneous view of virtualized resources [33]. While related work such as [10] and [5] use homogeneous setups, investigating the effect of heterogeneity could further improve the results in this paper and is part of our ongoing work.…”

Section: Threats To Validitymentioning

confidence: 97%

“…In this formalism, tasks are modeled as vertices, and precedence constraints are modeled as edges. DAGs are used across a variety of application domains, in science [9], industrial processes [10], and engineering [11]. DAGs also fit well web hosting [8] and multi-tier web applications [16].…”

Section: B Workflowsmentioning

confidence: 99%

“…Workflows are increasingly more common to use in datacenter environments, and are still rising in popularity across a variety of domains [5], [8]. Importantly, the characteristics of workflows vary widely across application domain, as indicated by the workflow size and per-task runtime reported for scientific [9], industrial processes [10], and engineering domains [11].…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

A Trace-Based Performance Study of Autoscaling Workloads of Workflows in Datacenters

Versluis

Neacsu

Iosup

2018

2018 18th IEEE/ACM International Symposium on Cluster, Cloud and Grid Computing (CCGRID)

Self Cite

View full text Add to dashboard Cite

To improve customer experience, datacenter operators offer support for simplifying application and resource management. For example, running workloads of workflows on behalf of customers is desirable, but requires increasingly more sophisticated autoscaling policies, that is, policies that dynamically provision resources for the customer. Although selecting and tuning autoscaling policies is a challenging task for datacenter operators, so far relatively few studies investigate the performance of autoscaling for workloads of workflows. Complementing previous knowledge, in this work we propose the first comprehensive performance study in the field. Using trace-based simulation, we compare state-of-the-art autoscaling policies across multiple application domains, workload arrival patterns (e.g., burstiness), and system utilization levels. We further investigate the interplay between autoscaling and regular allocation policies, and the complexity cost of autoscaling. Our quantitative study focuses not only on traditional performance metrics and on state-of-the-art elasticity metrics, but also on timeand memory-related autoscaling-complexity metrics. Our main results give strong and quantitative evidence about previously unreported operational behavior, for example, that autoscaling policies perform differently across application domains and by how much they differ.

show abstract

Section: Threats To Validitymentioning

confidence: 97%

Section: B Workflowsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

A Trace-Based Performance Study of Autoscaling Workloads of Workflows in Datacenters

Versluis

Neacsu

Iosup

2018

2018 18th IEEE/ACM International Symposium on Cluster, Cloud and Grid Computing (CCGRID)

Self Cite

View full text Add to dashboard Cite

show abstract

“…First, we expect that privacy and other data regulations, such as GDPR, 6 will increasingly affect the performance characteristics of cloud computing. For serverless computing specifically, the dynamic and ephemeral nature makes it an appropriate model for privacy-sensitive workloads, allowing platforms to schedule the execution on a granular level.…”

Section: Ideas For the Longer-term Futurementioning

confidence: 99%

“…For the workload we used the 10-minute Chronos trace [6]. This workload trace is an ETL workload, which originates from an industrial process use case.…”

Section: Event Propagationmentioning

confidence: 99%

Beyond Microbenchmarks: The SPEC-RG Vision for a Comprehensive Serverless Benchmark

Eyk

Scheuner

Eismann

et al. 2020

Companion of the ACM/SPEC International Conference on Performance Engineering

Self Cite

View full text Add to dashboard Cite

Serverless computing services, such as Function-as-a-Service (FaaS), hold the attractive promise of a high level of abstraction and high performance, combined with the minimization of operational logic. Several large ecosystems of serverless platforms, both open-and closed-source, aim to realize this promise. Consequently, a lucrative market has emerged. However, the performance trade-offs of these systems are not well-understood. Moreover, it is exactly the high level of abstraction and the opaqueness of the operational-side that make performance evaluation studies of serverless platforms challenging. Learning from the history of IT platforms, we argue that a benchmark for serverless platforms could help address this challenge. We envision a comprehensive serverless benchmark, which we contrast to the narrow focus of prior work in this area. We argue that a comprehensive benchmark will need to take into account more than just runtime overhead, and include notions of cost, realistic workloads, more (open-source) platforms, and cloud integrations. Finally, we show through preliminary real-world experiments how such a benchmark can help compare the performance overhead when running a serverless workload on state-of-the-art platforms.

show abstract

Smart Entrepreneurial Systems: An Application of Deep Reinforcement Learning in Improving Entrepreneurship Mentorship

Abhari

Williams

Pawar

et al. 2021

Advances in Intelligent Systems and Computing

View full text Add to dashboard Cite

Ananke: A Q-Learning-Based Portfolio Scheduler for Complex Industrial Workflows

Cited by 6 publications

References 38 publications

A Trace-Based Performance Study of Autoscaling Workloads of Workflows in Datacenters

A Trace-Based Performance Study of Autoscaling Workloads of Workflows in Datacenters

Beyond Microbenchmarks: The SPEC-RG Vision for a Comprehensive Serverless Benchmark

Smart Entrepreneurial Systems: An Application of Deep Reinforcement Learning in Improving Entrepreneurship Mentorship

Contact Info

Product

Resources

About