Wednesday, February 13, 2013

Best Practices - Data Center Upgrades Without the Downtime

 
The data center reflects the fortunes of the business it supports and, as such, has its own cycle of growth and consolidation. Greater usage and data throughput means more servers and racks; this, in turn, means more heat and power issues, which inevitably lead to a need for upgrades or a consolidation of the center itself. Add to this the emerging environmental and energy-consumption regulations, and it is easy to understand why data center managers find themselves in a state of almost perpetual change. Although a new, purpose-built facility always offers an attractive solution to accommodating these demands, it is in fact rarely an option. Because of the cost and disruption issues associated with new construction, retrofitting and live upgrades account for the vast majority of change in the data center, outpacing new construction by a ratio of around three to one.
So how do you prepare for a successful upgrade without downtime?
 

Think About the Business, Not the IT

 
The first, and most important step is to understand that live upgrade is a business issue rather than an IT issue. The choice of a live upgrade is driven by demands on the business to minimize cost and disruption. Therefore, in this context aligning the planning and preparation for the live upgrade to the business itself is of paramount importance.
This approach is very much in the interests of the business as well. A quick checklist of the risks to core business processes that could stem from a failure in the data center makes a convincing case: communications, email, customer accounts and services, the maintenance of an auditable data trail, and many others should convince any board that a live upgrade is a crucial business project.

Properly Supply the Project

Establishing the live upgrade as something that is not “business as usual” should enable you to avoid the most common mistake made by organizations considering such projects: making the IT department wholly responsible for its planning and execution. To make this mistake represents a failure to understand that an IT department’s inherent strength is its consistency and resilience as a service provider. It does not necessarily follow that the IT department should also offer, from its existing resources, the project and change-management skills that are absolutely essential for a successful upgrade. A live upgrade requires a lot of planning across the business. In fact, the project is usually 80% preparation and 20% execution, and as one CTO said, “If my staff had the time to plan a data center upgrade, then I’d know I was overstaffed!”
 
There are four distinct phases of preparation that must always precede the upgrade itself.
 
Phase One: Communicate
 
Although a properly functioning data center rarely has many visitors, it does have many stakeholders. These stakeholders comprise the various parts of the business that depend on the data center to support their processes, and when you begin to contemplate an upgrade, you must start with a communications program that engages with all of them. The program should also include the contractors, third-party support organizations, the project team and any others who hold responsibility for the data center environment. Identify your stakeholders, tell them how the upgrade project will affect them and win their support.
 
Phase Two: Understand Your Environment
 
A thorough audit of your IT environment will help you gain a deep understanding of the demands of the project. Different technologies have different requirements, so you should not assume one size fits all—for example, hot and cold aisles work well for some servers, but not if they draw air from the side. Some equipment, such as a SAN, is very heavy, so make sure you know what the requirements are and plan accordingly. Similarly, consider cable lengths when planning the equipment layout.
 
Phase Three: Plan the plan
 
For the project to be successful, you need to take the business with you. This means listening and understanding the challenges that your colleagues face across the business and planning the work in manageable chunks to suit their needs as well as your own. As you move from planning to execution, establish site control and ensure there is a statement of work (SOW) for each activity. The best approach is to regularly test and validate the plan and be prepared to compromise.
 
Phase Four: Mitigate the Risk
 
Impact assessments for each element of the work are an invaluable tool for managing risk throughout the project. These assessments should anticipate the effects of the work itself and plan for the dust and rubbish that will inevitably be produced; failure here could present the biggest risk of all. A live upgrade is as much about change management as it is about IT, so if your business has a process you should use it. Additionally, ensure that the data center’s backup regime is reviewed before embarking on the project. This should include a review of its physical and logical security as well and the business continuity plan—after all, the data center is about to become a building site!
 
In conclusion, the live upgrade of a data center is not a business-as-usual activity; a typical project may last 6 to 12 months. The data center must support the business for 6 to 12 years. The years between upgrades will involve significant changes both in technology and best practice, so accessing external resources for the project brings the benefit of state-of-the-art specialist knowledge as well as the additional people needed to get the job done on schedule. External expertise—whether consultants or contractors—should offer project-management skills, specialist knowledge, and data center operations experience as well; after all you don’t want them to deliver you a newly refurbished data center that is impossible to operate!
 

No comments:

Post a Comment