In my previous article, we examined the business drivers for a cloud migration. Next, let’s look at some strategies for success, and some common challenges when moving and IT organization to the cloud.
There isn’t just one path to the cloud. When planning migrations with my clients, I present a number of proven options, with a focus on not only minimizing downtime, but also a path to growth. Most migration plans fall into one of three categories: Forklift, Refactor, or Rebuild. Let’s dive into each…
The forklift strategy is generally the most popular. This is the least complex, as it simply involves copying the existing software as-is to a cloud service provider, often to a simple Virtual Private Server (VPS). None of the scalability and elasticity of the new infrastructure is leveraged, and this is essentially virtual colocation. Keep in mind that differences in the cloud environment will require attention to some areas such as network connectivity, authentication, and monitoring.
Many of my clients select this option simply due to lack of internal resources or experience rearchitecting applications for a more sophisticated cloud environment. I often help clients design and plan for elasticity, security, and high availability in a cloud infrastructure.
Refactoring involves programmatic changes to software in order to take advantage of Infrastructure-as-a-Service (IaaS) or Platform-as-a-Service (PaaS) cloud environments. The most common application changes I make with clients involve application state. Very often, applications will hold client session data in local memory, making it impossible for subsequent requests to visit another server in the environment. This prevents both scalability and availability.
For large-scale e-commerce applications, these concerns becomes critical, as adding front-end web server instances improves response times, as well as scalability and availability. The business case for this refactoring is clear; minimizing response times improve not only search engine rankings, but also conversion rates.
Other refactorings might involve the integration of a Content Distribution Network (CDN), offered by some cloud service providers. Amazon CloudFront, Akamai, EdgeCast, and Limelight are some examples. A CDN will allow applications to serve content, such as static web pages and images, to users with greater performance and reliability. It can also greatly reduce incremental bandwidth usage fees.
Many legacy applications simply don’t port well to the cloud. I’ve worked with many companies whose custom line-of-business applications are too archaic to move. Some might rely on an unsupported operating system, application framework, or other defunct technology. Others might require integration with other on-premise systems that also don’t easily move to the cloud. These applications are candidates for a wholesale rewrite.
While a brand new cloud-native system might sound attractive, there are some potential risks and disadvantages to consider. The primary concern here is the additional capital costs involved in rewriting an application from scratch. Sometimes existing application components can be reused, but these are typically limited to business logic, and are usually thrown away in deference to newer technologies.
Another risk is vendor lock-in. Choosing cloud-native infrastructure and services will serve you well, but I caution using vendor-proprietary services. I usually recommend open source solutions and services that are available across vendors. This allows an easier migration between cloud service providers, in order to take advantage of future innovations that one provider may not yet offer. Also, given the “here today, gone tomorrow” nature of the industry, selecting technology proprietary to a specific cloud vendor is a risk to consider avoiding.
Finally, consider the organizational changes a migration to the cloud may necessitate. For some, this may involve training to acquire new skills. Individual positions may need to be eliminated, as legacy applications are replaced, and their maintenance becomes unnecessary. Some organizations will eliminate entire departments, save perhaps for a minimal support staff, as the on-premise datacenter is obsoleted.
The Cloud CIO
As the need for on-premise infrastructure management wanes, CIOs will also shift their focus from implementer to vendor manager. This could be a welcome change for many CIOs, who often long for more time to enjoy their role as a business leader and strategist.
In your role as a business leader, how are you using the cloud to create value?