Cloud computing platforms are different from physical and virtualized computing platforms. It is essential to understand the limitation and opportunities while creating solutions for a cloud environment. It also entails the need to look again at architecting methodology. Here we highlight steps you can follow while architecting for public cloud deployment.
Step 1 :
Develop a modularized functional architecture for the cloud solution, including the underlying platform components needed.
Step 2 :
Develop the operational model for the cloud solution functional components (platform and infrastructure services needed).
- Determine computing resource requirements for the functional components (including their respective platform components)
- Determine backup requirements.
- Determine system management requirements.
- Determine scalability and high availability requirements for each of components.
- Determine performance requirements, probable bottelneck and failure points of each of the functional components.
Perform a fit-gap analysis for the functional components against the target cloud environment. This will require conversion of physical compute resource values to requirement on cloud environment and may require knowledge of Relative Performance Estimate 2 (RPE2) values or similier benchmarks. At this stage you also identify resources that may not fit on the cloud as-is—such as large CPU or storage requirements.
Based on results of Step 3, scalability and high availability requirements as identified in step 2, re-factor the solution and deployment architecture to enable large compute requirements on the cloud. This may require parallel, de-coupled, elastic and multi-site deployment architecture.
Validate the architecture against functional and non-functional requirements.
Determine the cost of hosting the solution.