Management Services for
Cloud Computing

Cloud computing is a new deployment and operational model in which high level computation services and storage are provided by Internet ("cloud"). In this project we investigate a two layered cloud computing model: in the first layer, a federation of sub-clouds offers basic infrastructure services such as storage and raw computation; the second layer offers higher level computational services that provision and manage software development platforms, including Tools as a Service. The project addresses the challenges in automating the cloud management and provides services for cloud federation, brokerage of resources, optimization and performance prediction.

This project won the “IBM CAS Project of the Year Award.”

For more information on this project, click here.

This project was again highlighted during the Innovation Impact session at CASCON 2011. Below is a video presentation of this more recent work on:

Multi-model Adaptive Cloud Environments (MACE)
  • One advantage of a cloud offering is its elasticity (i.e., it may grow and shrink its footprint in response to fluctuations in demand). Typically, sets of rules are defined which govern this elastic behaviour. An alternative to rules is the use of an optimisation model and feedback loop. This can provide precise and accurate predictions of resource needs and result in improved adaptive behaviour. Both approaches have their benefits and limitations. A Multi-model Adaptive Cloud Environment (MACE) is introduced in which a cloud manager utilizes a hybrid approach, actively switching among models as needed, to drive its provisioning decisions.

Innovation Impact: Multi-model Adaptive Cloud Environments (MACE) from IBM Canada CAS on Vimeo.

Technological Services
Representative Papers 

[1] Hamoun Ghanbari, Bradley Simmons, Marin Litoiu, Gabriel Iszlai: Exploring Alternative Approaches to Implement an Elasticity Policy. IEEE CLOUD 2011: 716-723,

[2] Cornel Barna, Marin Litoiu, Hamoun Ghanbari: Model-based performance testing. ICSE 2011: 872-875.

[3]Cornel Barna, Marin Litoiu and Hamoun Ghanbari:  Autonomic load-testing framework. ICAC 2011: 91-100.

[4] Hamoun Ghanbari, Bradley Simmons, Marin Litoiu, Gabriel Iszlai: Feedback-based optimization of a private cloud. Future Generation Comp. Syst. 28(1): 104-111 (2012).

[5] Bradley Simmons, Hamoun Ghanbari, Marin Litoiu, Gabriel Iszlai: Managing a SaaS application in the cloud using PaaS policy sets and a strategy-tree. CNSM 2011: 1-5.

[6] Bradley Simmons, Hamoun Ghanbari, Sotirios Liaskos, Marin Litoiu and Gabriel Iszlai: “Software Engineering for Self- Adaptive Systems 2.  Chapter. Hierarchical Self-Optimization of SaaS Applications in Clouds.  354-375 (2013).

[7] Hamoun Ghanbari, Bradley Simmons, Marin Litoiu, Cornel Barna, Gabriel Iszlai:  Optimal Autoscaling in the IaaS Cloud.  ICAC 2012.  173-178.

[8] Mike Smit, Bradley Simmons and Marin Litoiu: Monitoring Next-Generation Cloud Systems.  Future Generation Computer SystemsTo appear 2013. 17 pages.

[9] Mike Smit, Przemyslaw Pawluk, Bradley Simmons and Marin Litoiu:  A Web Service for Cloud Metadata.  SERVICES 2012:  361-368.    

[10] Przemyslaw Pawluk, Bradley Simmons, Mike Smit, Marin Litoiu and Serge Mankovski: Introducing STRATOS:  A Cloud Broker Service.  CLOUD 2012:  891-898.

[11] Mike Smit, Mark Shtern, Bradley Simmons and Marin Litoiu: Partitioning Applications for Hybrid and Federated Clouds.  CASCON 2012:  27-41.

[12] Mark Shtern, Bradley Simmons, Mike Smit and Marin Litoiu: An Architecture for Overlaying Private Clouds on Public Providers.  CNSM 2012:  371-377

[13] Mark Shtern, Bradley Simmons, Mike Smit and Marin Litoiu:  Navigating the Clouds with a MAPAccepted to appear. The 13th IFIP/IEEE IM 2013: 1-7.