Assisted migration of enterprise applications to the Cloud – A hybrid Cloud approach

Abstract

Cloud Computing is a relatively new paradigm with the potential to transform how IT
hardware and software are designed and purchased. Computing is no longer purchased
as typical products but delivered as a service over the Internet from large data centres.
However, despite the potential benefits associated with the migration of enterprise applications from an in-house data centre into a Cloud infrastructure, there are still some
issues that hinder the process. In this respect, a migrated application should satisfy specific enterprise policies related to privacy and security, as well as provide an acceptable
quality of service. Because of the complexity of enterprise applications today with respect
to the large number of components deployed in multi-tier architectures, the complexity of
the interactions between them and the components’ dependence on stored data, the migration process has to be assisted. Therefore it is presented a framework able to automate the decision-making process related to the migration of different components to a virtualized Cloud system based on a model which factors in benefits and disadvantages
associated with the migration subject to the fulfilment of the SLA and the respect of policy
constraints. The presented approach envisions an application hosted both partly onpremise and on the Cloud. Resulting migrated applications need to be compliant with
enterprise and security policies’ constraints while minimizing costs and ensuring performance in terms of wide area network communications and response times. The framework relies on a model to efficiently decide on the component placement. A model which takes into account several factors such as enterprise policies, dynamic performance bottlenecks, data sensibility, performance, cost savings from migration, data flows between application components, and the spread and variability of users. On the one hand the migration could lower the SLA (Service Level-Agreement) while on the other hand the
deployment on the Cloud would provide the ability to handle peaks in workload and a
higher reliability due to the higher level of replication, the existence of multiple fault
domains and the deployment of the application on multiple Clouds. With respect to the
latter, it will be assessed the deployment on different Clouds depending on the constraints
imposed by them. The presented framework relies on algorithms and a formal decision
model in order to assist developers in migrating existing applications. With the aim of
empirically evaluate the overall framework, a prototypical implementation of the framework will be used in order to support the migration to a Cloud-based environment of an existing product configurator. The empirical validation will be used to further improve
the input model on which the decision-making process was founded.

PDF

This entry was posted in Workshops. Bookmark the permalink.

Comments are closed.