Microesrvice Resource Autoscaler

Last updated on Oct 19, 2022
Code
The block diagram of our system

A resource manager on top of kubernetes that automatically finds optimum resources for deployed application without human intervention. Upon doing that, the resource manager does not violate SLO and does not need offline datasets for training. The algorithm is simple yet effective in real world scenario.

Results: Saves 33% more resources than state-of-the-art rule based autoscaling Technology used: Python, Kubernetes, Prometheus, Jaeger, Machine Learning.

Resource Autoscaler Microesrvice Kubernetes
Md Rajib Hossen
Md Rajib Hossen
PhD Candidate in Computer Science

My research interests include Microservices, HPC, Distributed Systems, Machine Learning, and Converged Computing

Publications

Practical Efficient Microservice Autoscaling with QoS Assurance
In this paper, we propose PEMA (Practical Efficient Microservice Autoscaling), a lightweight microservice resource manager that finds efficient resource allocation through opportunistic resource reduction. PEMA’s lightweight design enables novel workload-aware and adaptive resource management.
Md Rajib Hossen, Mohammad A. Islam, Kishwar Ahmed
PDF Cite Code Project Slides DOI
Practical Efficient Microservice Autoscaling
In this paper, we propose PEMA (Practical Efficient Microservice Autoscaling), a lightweight microservice resource manager that finds efficient resource allocation through opportunistic resource reduction. PEMA’s lightweight design enables novel workload-aware and adaptive resource management.
Md Rajib Hossen, Mohammad A. Islam
Code Project Poster
Towards Efficient Microservices Management Through Opportunistic Resource Reduction
In our work, we are developing a lightweight learning-based resource manager for microservices that does not require extensive data and avoid causing service level objective violation during learning. We start with ample resource allocation for microservices and identify resource reduction opportunities to gradually decrease the resource to efficient allocation.
Md Rajib Hossen, Mohammad A. Islam
PDF Cite Project Slides Video Source Document