Javascript must be enabled for the correct page display

Reducing operational costs of microservices by means of a deployment tuner

van der Knaap, Thijs (2019) Reducing operational costs of microservices by means of a deployment tuner. Master's Thesis / Essay, Computing Science.


Download (2MB) | Preview
[img] Text
Restricted to Registered users only

Download (141kB)


Running a microservices architecture on the cloud can have significant cost advantages, especially when load on the system varies. Operational cost can be minimised by ensuring that just enough resources are requested, without harming performance. Autoscalers have been developed which analyse the system and scale accordingly, but even more cost can be saved by applying deployment tuning. This work researches the financial benefits of expanding the standard Kubernetes autoscaler with an automated deployment tuner. The tuner reduces operational cost by actively searching for deployments that require less resources. The tuner is able to modify the current deployment and acts when the deployment is stable, this ensures that it does not compete with the autoscaler. It tunes based on the current load, deployment and stored previous deployments. When no cost gains can be made the tuner still expands its knowledge by performing small changes to the deployment, which also prevents the system getting stuck in a local minimum. The evaluation shows that the tuner decreases operational cost of a cluster consisting of a single microservice. These savings do come with a decrease in performance. Further evaluation with a full microservice system is needed to judge the full cost saving potential of deployment tuning.

Item Type: Thesis (Master's Thesis / Essay)
Supervisor nameSupervisor E mail
Degree programme: Computing Science
Thesis type: Master's Thesis / Essay
Language: English
Date Deposited: 30 Aug 2019
Last Modified: 11 Sep 2019 09:01

Actions (login required)

View Item View Item