Speaker:
Jairo A. Gutiérrez and William Liu

Date(s):
02 Sep 2012 to 02 Sep 2012

Topic:

Cloud Computing has emerged as a significant option for organisations to enjoy up-to-date, well supported and scalable computing facilities.  Several large, medium and even small organisations have adopted the technology and shifted their vision to use information technology as a service in order to reduce investment in infrastructure and complementary services. At the same time there are concerns about the energy impact of cloud-based solutions and on the trade-offs associated with energy efficiency versus enterprise-level performance.

This paper explores the suitability of the policy-based management (PBM) paradigm as a way of achieving the necessary balance between energy efficient cloud-based solutions and their performance.   A network endowed with autonomic management needs to address the resource allocation problem within the constraints of the business rules imposed on it: in this case energy efficiency while maintaining adequate performance levels. It also must own the means to learn and interpret changes in the surrounding environment; many authors refer to such means as policies, and their implementation as Policy-Based Management [1].  PBM proposes information and ontological modelling to capture knowledge relating to network capabilities, environmental constraints, and business goals and policies, together with reasoning and learning techniques, to enhance and evolve such knowledge [2].

While other authors have proposed an algorithmic approach to energy efficiency [3], the PBM approach is focused on providing support that permits to automatically assign network resources, control quality-of-service (QoS), and monitor resources, using a set of previously defined policies. A policy is a directive that manages certain aspects of the behaviour of the system and it depends on the interaction of users, services and existing resources. The implementation of the policies is based on the basic rule “if event and conditions then actions”. The flexibility of the PBM paradigm makes it a first candidate to service management in a dynamic and distributed context, not only because of the number of possibilities that could be opened but also because of their diversity [1].

In the scenario envisioned for this paper, the policies should be automatically generated without intervention of manual operators and derived directly from the requirements of the users, the services and the devices.

This paper contributes to understand how PBM can be used to specify, monitor and control cloud-based energy efficient solutions, at the infrastructure level, which are capable of providing dependable solutions for the upper cloud levels (PaaS and SaaS).

References

[1] F. Beltran, J. Gutierrez and J. L. Melus, "Technology and market conditions towards a new competitive landscape in the wireless access market", IEEE Communications Magazine, Vol. 48, No. 6, pp. 46-52, 2010.

 [2] B. Jennings, S. van der Meer, S. Balasubramaniam, D. Botvich, M. O Foghlu, W. Donnelly, J. Strassner, “Towards Autonomic Management of Communications Networks”, IEEE Communications Magazine, Vol. 45, No. 10,  pp. 112-121, 2007.

[3] S. Albers, “Energy-Efficient Algorithms”, Communications of the ACM, Vol. 53, No. 5, pp. 86-96, 2010

 

Download the Presentation [PowerPoint 2007 Slide Show - 1.98 MB]