Implementing Cloud Computing Takes a Complete Strategy
Editor’s note: For more on analysis, planning and implementation of a cloud computing project, see this related checklist.
It’s not too late to get started with cloud. Starting today allows you to exploit the wave of development and refinement that’s already taken place in software and service practice. Whenever you start, many decisions must be made in planning an in-house cloud implementation. You must consider myriad questions, which can be daunting.
- What hardware?
- How much virtualization?
- Which OSs are needed?
- What management software?
- What services like backup and restore, patching and security?
- What change windows are needed and when should they occur?
- What middleware should be supplied prebuild and standardized?
- Should labor services be added or kept as a one-of-a-kind option?
- Should packages be created for easy consumption like small, medium and large images sizes?
- Should infrastructure be set up for general use or built for a specific purpose like system development, analytics or desktop virtual workloads?
Consider the Following
Many considerations will influence the answers to these questions. For example, a purpose-built infrastructure is easier to plan and build than a general-purpose one. That’s because less technology selection is involved, thus a lower chance to miscalculate. Another consideration is standardization. A strong tension exists between standardization and the need for flexibility. However, too much flexibility will kill any hope of implementing cloud and saving money over business-as-usual solutions.
One way to get a handle on the cloud-implementation challenge is to break the effort into activities. Three key areas are analysis, planning and implementation. Each has factors to consider, which can be grouped into areas like benefits and risks, and technology scope. Figure 1 offers an overview of the steps to move forward and the things to consider. Let’s start with the analysis needed when working with cloud.
Your analysis’s main focus should be benefits and risks. The potential benefits include a pay-for-what-you-use pricing model, flexible scaling of computing resources, and rapid provisioning of images and related resources. After listing benefits, consider and assign potential impacts based on what you know about your company’s characteristics and potential for cloud adoption.
Potential risks are also important. These might include the exposure of servers and images to attack, loss of flexibility due to standardization, and complexity or uncertainty of flexible pricing. After listing such risks, consider and assign a potential impacts or mitigation. It’s important to do this considering what you know about your company’s culture and likelihood for cloud adoption.
The main focus of your planning activity should be developing an understanding of alternatives, technology scope and business model. Start with cloud alternatives.
What are your best options? Start with private and public infrastructure as a service (IaaS) and platform as a service (PaaS) options. For each, there are potential workloads to consider. Some typical workloads are shown in Table 1.
Table 1: Matching Workloads
|Development and Test
||General purpose private and public IaaS
||Purpose-built IaaS for set a of analytics applications
||Purpose-built IaaS for storage, backup, recovery, and archiving
||Purpose-built infrastructure for a set of collaboration applications
||Purpose-built infrastructure in support of desktop virtualization
||General purpose private and public IaaS
|Self-service application development environments
||Hybrid infrastructure and platform in support of application development
||Purpose-built infrastructure for email on the Web
|Database as a service
||Private or public PaaS supporting desired aspects of databases
|Middleware as a service
||Private or public PaaS supporting desired aspects of middleware
|Software lifecycle as a service
||Hybrid IaaS and PaaS in support of software design, development, and maintenance
What’s your best technology scope? Once you understand the potential alternatives and workloads, consider the scope of the technology you’ll deploy. You need to select specific technology and services, and these will be closely linked to the infrastructure and platforms you need. Make selections carefully because if you leave something out, it will have to be added later, and this will undoubtedly delay your project. Figure 2 lists three technology examples.
You’ll also need other categories, such as storage, OS and software images, security including policy, software and hardware, integration, and application programming interfaces. Remember, if something important is left out, it will have to be added later.
What’s the right business model? In addition to technology, consider the business model of your cloud implementation. This includes the pricing model featuring pay-as-you-go for customers of the service, dynamic scaling of resources, both up and down in size and capability, and service level agreements in areas like availability or provisioning time.
Teams must package features to make the solution a consumable cloud entity and the packages need to be useful and flexible. Designers might create a “large image” which contains capacity, software, human support and a specific service level agreement. How much of each feature should they include? The complexity of this challenge is shown in Figure 3.
These attributes have technology implications, but they’re mainly about the business qualities you intend to give to your cloud solution.
This last step involves many standard project-management practices. Begin by creating a detailed implementation plan for the service based on your planning and analysis steps. Next, procure the needed hardware and software. After their installation, test the hardware and software, as well as the new services built upon them—such as backup and recovery, and patching. As needed, you’ll have to revise the design, hardware, software and service.
In accordance with good project-management processes, pilot the overall solution before it goes into production. After you set up, test and go live, don’t forget to report on the success of the production implementation.
Is It That Simple?
Like many IT projects, implementing cloud will be a challenge. As you can see, there are many opportunities to make decisions. It’s important to document and explain each decision so an honest assessment can be made of the project after the first workload goes into production.