How to classify workloads for cloud migration and decide on a deployment model

This post was co-authored by Paul Alter

With enterprises aggressively enabling their data centers with cloud technology, they are looking to relocate applications to cloud for operational efficiency. For large enterprises, with hundreds of applications to be considered for migration, it is important to define a methodical approach including considering rationalization of the portfolio of applications that are to be migrated. They should, at a minimum, analyze the portfolio from the application technology perspective and classify them along with appropriate deployment model consideration. Having worked with a very large telecom enterprise and a large consumer electronics enterprise looking at hundreds of applications, I devised an approach that helps classify applications into different categories and then identify the appropriate deployment model. I will describe the classification approach and set of criteria to decide on the appropriate deployment model for those applications.

Categories of workloads for migration to cloud

Enterprises typically look at a portfolio of business applications from a functionality angle to slice and dice in an attempt to rationalize. However, cloud enablement brings a perspective that means they should look at the technological aspect of applications. Primarily, one should look at the application’s viability to run on a virtual machine stack and a commoditized and standardized platform as the key criteria. As depicted in the following diagram, the workloads and applications that are identified to be migrated into the cloud after the analysis may be categorized into:

Migrate through Lift and Shift. The process of workload migration through Lift and Shift includes making very minor changes to the workload. Such minor changes are done mainly to accommodate change of foundation technology platform components such as operating systems, network configurations and more. This category of migration of workloads is typically in the infrastructure as a service (IaaS) space, so the workload can run on a virtualized infrastructure environment. There typically are no changes to the business logic expected in this category of workload migration to cloud.

Transform. The Transformation category of workloads entails the process of application migration, including making changes to the application to enable the workload to run in a virtualized infrastructure and platform environment. The transformation of workload may require changes to the business logic to take advantage of the cloud environment which might include changes towards Services Oriented Architecture enabling loosely coupled application environments. Such changes may also help workloads and applications cater to API Economy models enabling APIs to be developed and externalized for better information access.

Rebuild. The Rebuild process of the application migration includes making significant changes to the application, like rewriting the application due the migration to the cloud. This usually occurs, as an example, when certain software libraries that are used in the application are no longer supported on the cloud such as existing monolithic applications. Also, the business logic organization will require significant changes to enhance the business capabilities.

A framework for selection of a deployment model for the migration of workloads to cloud

When workloads are migrated to cloud, enterprises often find it difficult to decide on the deployment model such as public, private or hybrid. There are several factors that play a role in the selection of appropriate deployment model for a workload to be moved to cloud. Such factors include security privacy regulation needs, unique technology needs for the workloads and agility and elasticity required of the workload. The following diagram depicts a high-level framework of decision criteria and matching appropriateness of specific deployment models.

Building on this framework, the IBM application Cloud Affinity Analysis (CAT) can provide a structured approach for assessing a given application. CAT assesses the business value relative in a given organization in migrating to a cloud against the ease or difficulty in making that migration, or pain versus gain. It also rates where on the continuum the application most likely sits against key cloud characteristics, such as private or public hosting, and IaaS to software as a service (SaaS) environments.

In this first graph, the size of the circle represents overall cloud affinity on a scale of one to ten. The pain in migrating the application to a cloud is affected by several factors such as the number and type of connections between the application and other applications, the amount and style (batch, interactive) of data transferred, the non-functional response characteristics for those data transfers, the security and compliance requirements associated with the application workload and several other factors. The gain is derived from aspects such as agility and elasticity requirements of the workload, value in rapid deployment of images, overall motivation of the organization to see this workload in a cloud and other characteristics.

The second graph shows the application affinity to IaaS to SaaS, and public or private cloud models. But it’s also important to realize that the application Cloud Affinity Analysis is an initial step in an end-to-end framework for identifying, analyzing, designing and implementing application migration to cloud. This framework will be futher explored in future Thoughts on Cloud postings.

I hope the classification model and the deployment selection criteria framework will be useful for those who consider migrating applications to cloud. Please feel free to share your views through comments below or send your questions to me through email.

Paul Alter is a Senior IT Architect for the Global cloud CoC. Paul has worked in application development, IT infrastructure, outsourcing, software product development, and IT consulting for over 32 years. His expertise includes application integration technologies and implementation, data center transformation and application migration, and application architecture. Working in the Global cloud CoC, Paul focuses on cloud strategy, architecture and migration.
Paul can be contacted at alterp@us.ibm.com

Share
Comments: 1
Ram Ravishankar

About Ram Ravishankar

Ram Ravishankar is the Chief Architect for the Global cloud CoC. Ram has over 20 years of experience in Information Technology and has been always in the fore front of advanced / emerging technologies. He is leading Enterprise Transformation to cloud initiative for Global cloud CoC within IBM helping some of the top Fortune 100 clients with transformational strategy and architecture to adopt cloud in the enterprises. His current focus is cloud Technology solutions including Migration, Integration, Enablement of API Economy and Internet of Things.
This entry was posted in Workloads and tagged , , . Bookmark the permalink.

One Response to How to classify workloads for cloud migration and decide on a deployment model

  1. Marco Prado says:

    Mr. Paul / Mr. Ram,
    I really liked your article, very educative, interesting and useful. The article in general was great but the CAT analysis brought my attention. Can you please tell me were can I get the theory of this analysis ?

Comments are closed.