Richard Joaquín Gil Martínez
Automated Planning to Support the Deployment and Management of
Applications in Cloud Environments
Thesis submitted for the PhD in Computer Science and Engineering, Instituto Superior Técnico
(IST), Universidade de Lisboa and Université
catholique de Louvain.
Abstract
Cloud computing has enabled myriad of applications to benefit from
dynamic provisioning of resources. These applications adapt their
configurations to satisfy business goals, when operating in dynamic
environments. Establishing techniques to govern the deployment and
management of resources in the cloud is a challenging task,
though. Cloud applications are in need of automated mechanisms for
decision-making that: (1) facilitate the exploration of large solution
spaces (defined by the combination of machine sizes, adaptation
actions and their effects, and the decision steps in the temporal
horizon), (2) can define deliberate plans to guide the system in a way
that satisfies requirements, optimizes performance and minimizes
costs, and (3) support the automated definition and revision of
policies to adapt the system under common conditions. Automated
planning, the task of finding a course of action from an initial state
to a desired state, offer opportunities to meet these needs and
improve the operation of applications in the cloud.
This thesis focuses on the design and evaluation of approaches that
exploit automated planning to support the deployment and management
of applications running in cloud environments.
To this purpose, three different techniques are proposed and evaluated
in the thesis, using real-life case studies of elastic scaling or
workflow execution. In more detail, this thesis presents: (1) a
solution to the (offline) generation of reactive policies that
exploits classical planning to support the definition and revision of
policies applicable under common conditions; (2) a solution to the
(online) generation of proactive plans, that uses temporal planning to
exploit the benefits of behavioral predictions for the definition of
long-term plans to re-configure interactive applications; (3) a
solution to (offline) generation of execution policies, that uses
probabilistic planning to tackle the uncertainty provoked by the
revocation of spot instance, in the deployment of workflow
applications with deadline constraints.
Selected Publications
- Automated Planning to Support the Deployment and
Management of Applications in Cloud Environments
- Richard Joaquín Gil Martínez
- PhD
Thesis. Departamento de Engenharia Informática, Instituto
Superior Técnico (IST), Universidade de Lisboa and Université catholique de
Louvain.
- March, 2019.
- Available pdf.
- Planning Workflow Executions when Using Spot
Instances in the Cloud.
- R. Gil, A. Lopes and
L. Rodrigues.
- Proceedings of the 34th ACM Symposium on Applied
Computing (Dependable, Adaptive, and Trustworthy Distributed Systems
Track), Limassol, Cyprus, April 2019.
- Learning Non-Deterministic Impact Models for
Adaptation.
- F. Duarte, R. Gil, P. Romano,
A. Lopes and L. Rodrigues.
- In Proceedings of the 13th
International Symposium on Software Engineering for Adaptive and
Self-Managing Systems (SEAMS), Gothenburg, Sweden, May 2018.
- AUGURE: Proactive Reconfiguration of Cloud
Applications using Heterogeneous Resources
- R. Gil, Z. Li, A. Lopes, L. Rodrigues
- In Proceedings of
the 16th IEEE International Symposium on Network Computing and
Applications (NCA 2017), Boston (MA), USA, October 2017.
- Automatic Generation of
Policies to Support Elastic Scaling in Cloud Environments.
- R. Gil, A. Lopes, and L. Rodrigues.
- In
Proceedings of the 32nd ACM/SIGAPP Symposium on Applied
Computing (SAC), Dependable and Adaptive Distributed Systems
Track, Marrakesh, Morocco, April 3-7, 2017.
Luís Rodrigues