Forecasting resource availability for supporting real-time container migration

Hard real-time applications have stringent response time requirements and failure to meet the deadlines can have catastrophic consequences. With the success of the edge computing paradigm, more edge applications have real-time requirements. User mobility and multi-tenancy on the edge nodes create new challenges for these real-time applications running inside containers. These problems can be resolved by migrating the container along with all of the underlying files to a new node where the application can run safely. However, single-target migration is prone to failure due to network issues or system crashes at the target. It is also hard to guarantee that applications will meet their real-time requirements during (or even after) the migration process. To this end, we propose a framework for planning reliable real-time container migration. This framework monitors the container and nodes in the network and accurately predicts the future resource availability of the nodes. The forecasts are used to identify a subset of nodes that have sufficient resources to run the container. We migrate to multiple nodes in parallel to make it resilient to network issues and crashes.

Name thesis:degree_name

M.S.

Level thesis:degree_level

Thesis

Discipline thesis:degree_discipline

Computer Science

Grantor

University of Illinois at Urbana-Champaign

Year dc:date

2022

Author dc:creator

• Suresh Babu, Vishakh

Contributors dc:contributor

• Mohan, Sibin

• Real-time
• Container
• Migration
• Forecasting

Statement dc:rights

• Copyright 2022 Vishakh Suresh Babu

Language dc:language

en, eng

Handle dc:identifier

https://hdl.handle.net/2142/116124

OAI record oai:identifier

https://www.ideals.illinois.edu/oai-pmh?verb=GetRecord&metadataPrefix=etdms&identifier=oai%3Awww.ideals.illinois.edu%3A2142%2F116124