Autonomic deployment decision making for big data analytics applications in the cloud

Autonomic deployment decision making for big data analytics applications in the cloud

Decision making

Big data

Constraint theory

Middleware

Cloud analytics

Constraint satisfaction problems

When changes happen to big data analytics (BDA) applications in the Cloud at runtime, the affected BDA applications have to be re-deployed to accommodate the changes. Deciding the most suitable deployment is critical and complicated. Although there have been various research studies working on BDA application management, autonomic deployment decision making is still an open research issue. This paper proposes a deployment decision making solution for BDA applications in the Cloud: first, we propose a novel language, named DepPolicy, to specify runtime deployment information as policies

second, we model the deployment decision making problem as a constraint programming problem using MiniZinc

third, we propose a decision making algorithm that can make different deployment decisions for different jobs in a way that maximises overall utility while satisfying all given constraints (e.g., cost limit)

fourth, we design and implement a decision making middleware, named DepWare, for BDA application deployment in the Cloud. The proposed solution is evaluated in terms of feasibility, functional correctness, performance and scalability. 2015, Springer-Verlag Berlin Heidelberg.

4501-4512

16

21

Lu, Qinghua

Li, Zheng

Zhang, Weishan

Yang, Laurence T.

Soft Computing

2017

journalArticle

14327643

10.1007/s00500-015-1945-5

http://dx.doi.org/10.1007/s00500-015-1945-5