Cloud Computing Wiki

Decision template for cloud computing

This article was provided by Subra Kumaraswamy

Perhaps a decision template that takes the following into
consideration will help (by no means this is a comprehensive list):

1. Work load profile
 a) Average workload requirements - CPU, Memory, Storage, bandwidth
 b) Peak workload requirements - Variance of  CPU, Memory, Storage,
bandwidth and time duration of sustaining peak loads
This will force the workload profiling requirement.

2. Architectural patterns and consideration
 a) If legacy app,  cost to move to cloud bursting model or to fully
hosted app in the cloud model
 b) If new app, integration requirements and cost to integrate with
another application running in private cloud.
 c) Elasticity support requirements - What new tools are required to
support workload orchestration? and the cost to deploy and mange the
tools.

3. Technology requirements
  a) OS, DB, App stack vendor or specific software requirement -
Provide solid evidence on these requirements. Add licensing cost based
on the licensed model.
  b) Load balancing between private and public cloud or between
disparate public clouds
  c) Integration of KPI (key performance indicators) of app running
on cloud with existing monitoring tools.

4. Security, privacy and compliance requirements (this will dictate
what new controls are required to meet the existing compliance and
security requirements)
   a) Cost to implement new controls such as encryption if required
   b) Identity and access management of users in cloud (if required)
   c) Cost to maintain existing controls in cloud E.g. log
monitoring, access monitoring, forensic evidence preservation,
separation of duties, patching

5. End user/Business user requirements
  a) Latency to connect to the application
  b) Frequency of information accessed - This will feed into
architectural requirements when consolidation information across many
database, logs, etc
  c) SLA requirements on availability and support

BTW, most companies may have an existing method to madness in
deploying applications and may have a rationalized documented process
to provision resources to applications. Fortune 1000 Companies that
are thinking of taking advantage of cloud service should have a
template that documents the requirements similar to the aforementioned
decision template . If not, IMHO, the success may not be guaranteed.

Startups looking to leverage clouds are a whole different beast
because they are measured by time to market more than anything else.
Hence they value elasticity (cloud bursting) more than any other
virtues of cloud computing and they are not burdened by issues such as
legacy app integration, extending existing IT processes to cloud,
security and compliance requirements that are faced by established
companies.