The cloud: avoiding provider lock-in

Enterprises want the economics and elasticity of cloud computing, but are understandably reluctant to completely abandon their own data centres, or lock themselves into particular cloud architectures.

The perception is that, once your environment has been embedded into a service provider’s data centre (a hybrid cloud), it becomes impossible to switch to another service provider in the future without starting over.

These challenges must be addressed before applications can be successfully transferred:

– Application complexity:Enterprise applications commonly depend upon shared services, including directories and databases, residing in the enterprise data centre. These applications must, however, operate in the cloud without migrating or duplicating these shared services.

– IT silos:While the enterprise data centre and the cloud represent distinct infrastructure environments, care must be taken not to create duplicate tools, processes or teams.

– Secure, private connectivity:Incompatible network policies, unaligned IP address ranges and other tactical considerations must be overcome in order to allow applications to span on-premise and cloud environments.

Before an application can be transferred to the cloud, several things must be taken into account. For example, the application likely consists of multiple components, such as directory services, that are shared by other applications and cannot be easily migrated. Therefore, secure connectivity between the enterprise and cloud data centre must be established to use application components that still reside within the enterprise premise…

Cloud is also top of the agenda in a forthcoming CIO Executive Council Summit on February 9, featuring outgoing BA CIO Paul Coby as a speaker.

…In addition, there is more to the workload than just the core application stack. Networking and storage play an important role when determining the feasibility of migrating an application to a target cloud. The fundamental challenge is how to move application workloads without having to re-architect the complete application stack simply to adapt to entirely different storage strategies, networking architectures and security policies.

The answer is cloud-bridging. A cloud bridge makes hosted applications appear as though they are running within a single, contiguous enterprise network. This allows administrators and management to interoperate with the application as if it resides within the enterprise network. Applications no longer have to be re-architected or carved out from the existing data centre environment in order to function properly.

The bridge is critical to building a cloud-extended data centre whilst avoiding the many pitfalls that await a major cloud initiative. Migrated applications can seamlessly use common application services hosted in the enterprise data centre and reconfiguration or rewiring of the enterprise network is eliminated as a requirement. This simplifies on-boarding but also provides the all-important exit strategy should the customer choose to switch service provider.

Bridging is achieved by the encryption of all bi-directional data between the enterprise network and cloud infrastructure using industry-standard methods, such as IPSec to preserve security. In addition WAN optimisation technologies should be incorporated to overcome potential performance impairments due to distance.

Finally, the on-boarded application will require a holistic view of the entire workload. By including some type of graphical user interface (a web-based self-service portal, a console application or both) user experience over the cloud bridge becomes seamless.

A thoughtful strategy for tramsferring applications into the cloud will reduce both costs and risks to the enterprises’ business and avoid the painful process of re-architecting the application stack or the network infrastructure. It becomes a requisite step to realising better economics and the improved elasticity promised by cloud-based computing.

Pic: kevindooleycc2.0