Some Types of Applications Float Better Than Others | Gaining Altitude in the Cloud

Posted On June 27, 2011

Emerging delivery models and technology advances are altering the risk versus benefit tradeoff for the adoption of cloud computing. However, not all applications are good candidates for deployment to the cloud. Even if the characteristics of a particular application allow it to go to the cloud, the appropriate delivery model depends on the type of workload being deployed. Let’s take a look at how to identify the right delivery model for different workload types, and how to conduct a next generation IT assessment for your organization’s best-fit cloud computing model deployment.

To set the stage (although you’ve likely read this before,) there are multiple models for deployment of cloud services:

  • A Public Cloud is characterized by virtualized hardware and software applications that are shared by all users. The service provider runs and maintains this environment for the benefit of the subscriber clients, and the environment is accessed via the Internet. Amazon Web Services is a primary example.
  • A Private Cloud has virtualized hardware and software that is dedicated to a single client. This environment may be onsite with the client or maintained by a service provider in a collocation facility. Access to the environment is via the client’s corporate intranet. Rackspace is an example of a provider in this space.
  • A Hybrid Cloud is a combination of both environments. Think of a situation in which a company develops a private cloud for its baseline utilization but engages a public cloud provider to enable it to “burst” into the cloud during peak volume periods.

Virtualization, standardization and automated service management are all prerequisites for a cloud environment, regardless of the delivery model. Automated service management refers to an environment in which service catalogs, governance, provisioning logic and usage accounting/billing have been specifically architected for the cloud.

Workloads are defined by the various types of IT work the enterprise needs to perform. Examples of major workload types include:

  • Analytics/business intelligence
  • CRM
  • E-mail/messaging
  • ERP and supply chain
  • Industry/LOB applications
  • Communication/collaborations tools
  • Personal computing/productivity
  • Test and development environment
  • A multitude of infrastructure workloads

Workloads can be identified by how they are classified against several characteristics including business criticality, data security requirements, utilization patterns, and complexity. These characteristics determine the type of cloud delivery model that is most appropriate for a particular workload type. For example, workloads that are highly standardized are normally good candidates for a public cloud, while workloads that require a high degree of compliance, complexity, or data protection are typically better suited to a private cloud model.

Database and application-oriented workloads are good examples that are appropriate for private cloud deployment. To date, the most commonly deployed workloads on public clouds have been CRM applications and web portals for audio/video conferencing. Application streaming is an example of a workload type that has been widely deployed to both private and public cloud environments, as well as hybrids.

Some workload types – e.g., applications that require a high degree of customization, utilization measurement (for chargeback or billing purposes) or auditability – have not yet been targeted for deployment to any type of cloud computing model. In addition, applications that are not readily virtualized are, by definition, not easily deployed to the cloud, regardless of the delivery model.

Everest Group has identified seven steps to developing a next generation IT model utilizing cloud computing delivery models:

  1. Survey your current environment, identifying your current cost baseline, the workloads supported, the current delivery model(s) (traditional, virtualized, etc.), and to the greatest extent possible, the server utilization rates by workload and delivery type
  2. Assess and select potential workload types for migration to a cloud model
  3. Determine the appropriate delivery model for each workload you are considering
  4. Establish a preliminary architecture for service management and hardware configurations
  5. Assess the economics
  6. Finalize your solution
  7. Implement

The process I’ve outlined above may appear daunting, but rigor and detailed analysis are required if you want to develop a viable strategy that delivers both cost savings and performance improvement. Not to heap any more on your plate, but an assessment of the communication and change management strategies required to support the implementation of your cloud solution are also critical.

Good luck!


Learn more about Everest Group’s cloud transformation expertise.

Everest Group Executive Viewpoints icon Related Articles