Managing Vertical Memory Elasticity in Containers

Nicodemus, Carlos H.Z.; Boeres, Cristina; Rebello, Vinod E. F.

doi:10.1109/ucc48980.2020.00032

Cited by 3 publications

(3 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We present a small set of experiments with the aim of briefly illustrating the potential practical effectiveness of the proposed pipeline or delayed scheduling approach. With an eye on applying this technique across cloud virtualization services, in particular, container services and FaaS solutions, we have implemented our own containerbased Sequence Alignment service on top of IaaS VM instances which is managed by VEMoC [Nicodemus et al 2020]. One drawback of this vertical memory elasticity controller is that it employs container suspension to be able to maximize memory utilization.…”

Section: Experimental Evaluation On Amazon's Aws Ec2mentioning

confidence: 99%

“…If the memory allocated to a container becomes insufficient, their strategy suspends and checkpoints it, and then re-starts the container in another instance with sufficient memory. The work of [Nicodemus et al 2020] manages container memory allocations by dynamically adjusting memory limits in conjunction with strategies that combine pausing and preempting containers when short of resources. As a result, improvements in memory utilization, cloud costs, and job throughput can be provided in comparison to existing commercial orchestration tools.…”

Section: Related Workmentioning

confidence: 99%

“…This means the possibility exists that a host or instance may have insufficient resources to attend its allocated applications. To avoid resulting slowdowns or failures, and additional costs, tools must either migrate instances or suspend them in an attempt to aggressively maximize the utilization of the memory available [Nicodemus et al 2020].…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Making the most of what you pay for by delaying tasks to improve overall cloud instance performance

Sodré¹,

Boeres²,

Rebello³

2022

Anais Estendidos Do XXIII Simpósio Em Sistemas Computacionais De Alto Desempenho (SSCAD Estendido 2022)

Self Cite

View full text Add to dashboard Cite

Resource elasticity and server consolidation have long been among two of cloud computing’s most relevant management tools. Yet, exemplified with a scientific application use case, this paper highlights how judicious scheduling of tasks can help maximize resource utilization and improve performance and costs for both users and cloud providers. Developing an efficient cloud service for DNA sequence comparisons is adopted as a motivating use case. Using the bioinformatics tool MASA that finds an optimal pair-wise sequence alignment, we propose a model for co-scheduling multiple alignments on a single cloud instance. The resulting, practically optimal, non-preemptive schedule can effectively double the throughput of MASA-based sequence alignment workflows.

show abstract

Section: Experimental Evaluation On Amazon's Aws Ec2mentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Making the most of what you pay for by delaying tasks to improve overall cloud instance performance

Sodré¹,

Boeres²,

Rebello³

2022

Anais Estendidos Do XXIII Simpósio Em Sistemas Computacionais De Alto Desempenho (SSCAD Estendido 2022)

Self Cite

View full text Add to dashboard Cite

show abstract

Cost-efficient scheduling algorithms based on beetle antennae search for containerized applications in Kubernetes clouds

Shen²,

Li³

et al. 2023

J Supercomput

View full text Add to dashboard Cite

Extending OpenStack Monasca for Predictive Elasticity Control

Lanciano,

Galli,

Cucinotta

et al. 2024

Big Data Min. Anal.

View full text Add to dashboard Cite

Traditional auto-scaling approaches are conceived as reactive automations, typically triggered when predefined thresholds are breached by resource consumption metrics. Managing such rules at scale is cumbersome, especially when resources require non-negligible time to be instantiated. This paper introduces an architecture for predictive cloud operations, which enables orchestrators to apply time-series forecasting techniques to estimate the evolution of relevant metrics and take decisions based on the predicted state of the system. In this way, they can anticipate load peaks and trigger appropriate scaling actions in advance, such that new resources are available when needed. The proposed architecture is implemented in OpenStack, extending the monitoring capabilities of Monasca by injecting short-term forecasts of standard metrics. We use our architecture to implement predictive scaling policies leveraging on linear regression, autoregressive integrated moving average, feed-forward, and recurrent neural networks (RNN). Then, we evaluate their performance on a synthetic workload, comparing them to those of a traditional policy. To assess the ability of the different models to generalize to unseen patterns, we also evaluate them on traces from a real content delivery network (CDN) workload. In particular, the RNN model exhibites the best overall performance in terms of prediction error, observed client-side response latency, and forecasting overhead. The implementation of our architecture is open-source.

show abstract

Managing Vertical Memory Elasticity in Containers

Cited by 3 publications

References 14 publications

Making the most of what you pay for by delaying tasks to improve overall cloud instance performance

Making the most of what you pay for by delaying tasks to improve overall cloud instance performance

Cost-efficient scheduling algorithms based on beetle antennae search for containerized applications in Kubernetes clouds

Extending OpenStack Monasca for Predictive Elasticity Control

Contact Info

Product

Resources

About