Short name: Container
Long name: New containers scheduling approach in Kubernetes
Call: F4Fp-SME-COD191119 (see call details)
Proposal number: F4Fp-SME-COD191119-01
SUMMARY REMARKS & TESTBEDS
The idea of our project is to propose a new approach to optimize the scheduling of several containers submitted online by users in a private infrastructure of nodes i.e. a cloud platform. In the literature, several containers scheduling studies and frameworks are proposed as Docker SwarmKit and Kubernetes. However, the containers submitted by users are defined in multi-criteria contexts, such as the number of used CPUs and the used memory size. The state of each node in the cloud infrastructure is also defined in multi-criteria modes, such as the number of running containers, available CPUs and available memory.
The novelty of our scheduling approach is to choose for each submitted container the most appropriate node which has a good compromise between multi-criteria variables. To select the best node, we propose to use the Technique for the Order of Prioritisation by Similarity to Ideal Solution (TOPSIS) algorithm. TOPSIS is a decision analysis multi-criteria algorithm used with success to solve several problems.
To demonstrate the potential of our approach, we plan to implement our strategy in Kubernetes with a minimum of change in the code source to be used directly with the next version of kubernetes.
- Container experiment (download the poster)
- Review Container experiment | FEC7 (download the slides)