The benefits of grid computing for your business can be significant. However, it’s not for everyone; and like any other new concept, there are many pitfalls to avoid when implementing it.
This article provides some grid computing basics; and points out some common mistakes that can derail your implementation efforts, along with suggestions for remedying them.
For those of you not quite up to speed on the subject, here’s a brief overview of grid technology:
At the centre of any grid is the concept of pooling various resources such as computing, storage, archiving and application software — and eventually also knowledge — and using those resources for several different purposes in a shared fashion.
In its simplest form, called grid computing, a grid is used for either compute- or data-intensive applications. When an application is run, you do not know which server, storage array or archive is involved. This is much like switching on a light or turning on your oven — you draw the required electricity from the power grid, but you do not know where it comes from. Nor do you care, as long as the grid supplies you with the power you need at acceptable rates.
The virtualisation of computing resources has four important effects:
1. Grids are ideal infrastructures to support the ‘anywhere, anyhow, anytime’ usage pattern.
2. By sharing resources not only technical but also human resources, you can more effectively utilise them and reduce the total costs of your IT operations. The better means you have of achieving this virtualization — of managing many systems as one — the more cost-effective your operations will be.
3. You can improve the availability of your applications, because you have given up the one-to-one correspondence between an application and a server. When one server in a grid breaks down, all applications that were running on it are seamlessly moved to some other server in the grid.
4. You can improve the quality of your IT services, because you can ensure timely availability of the required capabilities and capacities. To achieve this, a layer of middleware manages and monitors all the applications and hardware resources, including the network that links the different components together.
Grids are, in general, classified according to whether they are built within a single data centre, across an entire enterprise, or shared by several different enterprises or even larger communities. Grids can also be classified as special-purpose or general-purpose.
According to the Gartner Hype Cycle, which estimates the maturity and the visibility of emerging technologies, internal grids are already climbing up the slope to the plateau of productivity, whereas general-purpose grids are still on the initial rise towards the peak of inflated expectations.
Introducing grid into the business
When moving in this direction, you can start small in the IT department with, for instance, a server consolidation, which may be considered the first step towards grid computing.
It should be remembered, however, that although grid technologies are still emerging and maturing, introducing a grid is not primarily a technical challenge. Grid computing will strongly affect your business on all levels: personnel, organizational, technical, and cultural.
Virtualisation of resources of all kinds is at the heart of grid computing. To achieve this, a paradigm shift in thinking is required at all levels: from ‘my resources’, ‘my services’ and ‘my knowledge’, to resources, services and knowledge as utilities which are not directly owned by some organizational unit.
The idea of owning something needs to be given up. As this very often goes along with a certain loss of power, which is implicitly expressed by the ownership or control of significant installations, this is not an easy task to perform.
Organizational boundaries need to be traversed and bridged within the enterprise and maybe even between different organizations, if your company is closely partnering with others or tightly integrated into a supply chain. Because of this horizontal orientation, cultural, cooperational and coordinational aspects are particularly important and will make the difference between the success or failure of your project.
Business drivers, common mistakes and how to avoid them
Introducing grid computing is clearly a strategic issue. Executive management needs to carefully decide whether to introduce it or not, and if so, to what extent.
Once the strategic decision is taken, a holistic planning and implementation of the transition process is required to make the project a success.
Throughout the entire project, continuous and active executive sponsoring is absolutely vital.
From a business perspective, the main grid project drivers are the need to better support business agility, the need to lower total cost of ownership on both the system and staffing side, and the fear that the ever-increasing complexity of IT infrastructures will soon become unmanageable.
All this is in line with today’s general wisdom that IT is an internal enterprise business service, and the trends to integrate and consolidate, as well as to build, rely more and more on true, open standards.
Introducing grid computing is not easy. Overcoming cultural, organizational and human resources barriers makes it a challenging and fault-prone task. As well, a number of conceptual and technical challenges need to be mastered.
Here are some pitfalls to look out for, together with suggestions to overcome them:
• Failure to provide adequate executive management support and motivation. For a successful grid project, permanent, strong executive sponsorship and continuous motivation of all people involved is imperative.
• The belief that grid computing automatically solves managerial problems with the current infrastructure. Grid is not a panacea. Before you start, the house needs to be in order.
• The belief that introducing grid computing is just a technical challenge. In general, grid computing will strongly affect your business on all levels: personnel, organizational, technical, and cultural. Virtualisation of resources of all kinds is at the heart of grid computing. Organizational and staffing changes are unavoidable and at the heart of the transition process.
• The misconception that tailored grid solutions are available off-the-shelf and can be introduced overnight. Grid computing solutions are not of a “one size fits all” nature. How grid computing should be set up critically depends on your industry sector and your business processes.
• Failure to plan for storage and disaster recovery. Storage and disaster recovery should be an integral part of the effort to introduce grid computing.
• Underestimating security issues. Security considerations should be given highest priority, especially if the grid solution comprises Internet components.
• Beginning with the end in mind. Introducing grid computing must start with the careful analysis of business processes and applications in use. A preconceived notion of the final grid infrastructure can be dangerous.
Benefits of grid computing
If you avoid common mistakes, the benefits of grid computing for your business can be immediate and tremendous.
You can address short-term IT needs on a just-in-time basis for activities which cannot be foreseen or planned a long time in advance.
You can temporarily create capabilities otherwise unfeasible, or capacities elsewhere not available, to reduce design times, design defects or time-to-market for your products.
You can grow yo