Runtime Resource Management with Workload Prediction

Niknafs, Mina; Ukhov, Ivan; Eles, Petru; Peng, Zebo

doi:10.1145/3316781.3317902

Cited by 10 publications

(8 citation statements)

References 11 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Algorithm 1 describes our heuristic to solve this. It generalizes the solution in [8] for multi-threaded applications.…”

Section: System Model and Problem Definitionmentioning

confidence: 96%

“…However, these algorithms assume that applications are constantly running, and do not consider application reconfiguration. As mentioned in Section I, Niknafs et al [8] do consider application reconfiguration, but limited to single-threaded applications.…”

Section: Related Workmentioning

confidence: 99%

“…Recently, Niknafs et al [8] attempted to enlarge the scope of analysis for firm real-time applications by computing joint mappings. In the generated schedules, different applications might be mapped to the same core and executed according to the earliest deadline first (EDF) policy.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Energy-efficient Runtime Resource Management for Adaptable Multi-application Mapping

Khasanov

Castrillón

2020

2020 Design, Automation &Amp; Test in Europe Conference &Amp; Exhibition (DATE)

View full text Add to dashboard Cite

Modern embedded computing platforms consist of a high amount of heterogeneous resources, which allows executing multiple applications on a single device. The number of running application on the system varies with time and so does the amount of available resources. This has considerably increased the complexity of analysis and optimization algorithms for runtime mapping of firm real-time applications. To reduce the runtime overhead, researchers have proposed to pre-compute partial mappings at compile time and have the runtime efficiently compute the final mapping. However, most existing solutions only compute a fixed mapping for a given set of running applications, and the mapping is defined for the entire duration of the workload execution. In this work we allow applications to adapt to the amount of available resources by using mapping segments. This way, applications may switch between different configurations with varied degree of parallelism. We present a runtime manager for firm real-time applications that generates such mapping segments based on partial solutions and aims at minimizing the overall energy consumption without deadline violations. The proposed algorithm outperforms the state-of-theart approaches on the overall energy consumption by up to 13% while incurring an order of magnitude less scheduling overhead.

show abstract

“…Algorithm 1 describes our heuristic to solve this. It generalizes the solution in [8] for multi-threaded applications.…”

Section: System Model and Problem Definitionmentioning

confidence: 96%

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Energy-efficient Runtime Resource Management for Adaptable Multi-application Mapping

Khasanov

Castrillón

2020

2020 Design, Automation &Amp; Test in Europe Conference &Amp; Exhibition (DATE)

View full text Add to dashboard Cite

show abstract

“…When constraints e.g. on deadlines, are loose, the improvements are less significant [116]. The prediction techniques and configurations should be set carefully to fit with any cloud problem.…”

Section: A Prediction Accuracy Of Incoming Workloadsmentioning

confidence: 99%

Load Balancing and Server Consolidation in Cloud Computing Environments: A Meta-Study

Ala’anzy

Othman

2019

IEEE Access

View full text Add to dashboard Cite

The data-center is considered the heart of cloud computing. Recently, the growing demand for cloud computing services has caused a growing load on data centers. In terms of system behavior and workload, patterns of cloud computing are very dynamic; and that might serve to imbalance the load among data center resources. Eventually, some data-center resources could come to be over-loaded/under-loaded, which leads to an increase in energy consumption in addition to decreased functioning and wastage of resources. Just considering energy-efficiency (that can be attained efficiently by consolidate the servers) may not be enough for real applications because it may cause problems such as unbalanced load for each Physical Machine (PM). Therefore, this paper surveys published load balancing algorithms that achieved by server consolidation via a meta-analysis. Load balancing with server consolidation enriches the exploitation of resource utilization and can enhance Quality of Service (QoS) metrics, since data-centers and their applications are increasing exponentially. This meta-study, reviews the literature on load balancing and server consolidation and presents a ready reference taxonomy on the most efficient algorithms that achieve load balancing and server consolidation. This work attempts to present a taxonomy with a new classification for load balancing and server consolidation, such as migration overhead, hardware threshold, network traffic, and reliability.

show abstract

“…Machine Learning approaches [6,12,17,25,33] for runtime resource management have gained lots of traction in the past few years, especially in management of high performance systems and cloud servers. These methods need special tailoring before deployment on embedded and real-time systems in order to reduce their high computational overhead at runtime [32,50]. Specifically, machine learning techniques have been a promising trend for modeling the complexity of interaction among different on-chip resources and the corresponding effect on resource metrics [24].…”

Section: Related Workmentioning

confidence: 99%

Hessle-Free

Moazzemi

Maity

Rahmani

et al. 2019

ACM Trans. Embed. Comput. Syst.

View full text Add to dashboard Cite

As computing platforms increasingly embrace heterogeneity, runtime resource managers need to efficiently, dynamically, and robustly manage shared resources (e.g., cores, power budgets, memory bandwidth). To address the complexities in heterogeneous systems, state-of-the-art techniques that use heuristics or machine learning have been proposed. On the other hand, conventional control theory can be used for formal guarantees, but may face unmanageable complexity for modeling system dynamics of complex heterogeneous systems. We address this challenge through HESSLE-FREE (Heterogeneous Systems Leveraging Fuzzy Control for Runtime Resource Management): an approach leveraging fuzzy control theory that combines the strengths of classical control theory together with heuristics to form a light-weight, agile, and efficient runtime resource manager for heterogeneous systems. We demonstrate the efficacy of HESSLE-FREE executing on a NVIDIA Jetson TX2 platform (containing a heterogeneous multi-processor with a GPU) to show that HESSLE-FREE: 1) provides opportunity for optimization in the controller and stability analysis to enhance the confidence in the reliability of the system; 2) coordinates heterogeneous compute units to achieve desired objectives (e.g., QoS, optimal power references, FPS) efficiently and with lower complexity , and 3) eases the burden of system specification.

show abstract

Runtime Resource Management with Workload Prediction

Cited by 10 publications

References 11 publications

Energy-efficient Runtime Resource Management for Adaptable Multi-application Mapping

Energy-efficient Runtime Resource Management for Adaptable Multi-application Mapping

Load Balancing and Server Consolidation in Cloud Computing Environments: A Meta-Study

Hessle-Free

Contact Info

Product

Resources

About