In the previous Tech Talk series, we looked at the Network Computer. In this series, we look at what the network computer connects to a centralised computing platform, which provides not just the processing power to run the applications, but also the storage capability. It is exactly what mainframes and mini-computers did, till the era of client-server computing came into vogue. The horsepower was in the backend with the terminals providing the input-output capability. When the desktops took over, a significant portion of the processing moved to the client side. Then, came the Internet with its browser based front-end, and the talk of network computers running web browsers providing all the utilities that users needed.
That hasn’t exactly happened. But, as networks become more powerful and the focus shifts to not only the next users of computing but also addressing the twin issues of cost and complexity in managing desktops, the focus is again shifting to the benefits of centralised computing.
Much of the current discussion around server-centric computing has focused around leveraging resources which are deployed across the network and making them work together as a single large computer a computing grid. This approach makes a lot of sense both for applications which need a lot of power which no single computing cluster can easily provide and in enterprises where there is are distributed, under-utilised resources.
The computing grid that I will discuss later in this series is slightly different from both of these. The grid that I am thinking of to complement the network computers is a public computing grid which provides virtual desktops to network computers. It is not about aggregating a collection of existing resources from across the network. Instead, it is about creating a scalable and reliable platform to address the needs for potentially millions of users. It is a platform because it allows other independent software vendors to deploy their applications on this computing foundation. It offers the ability to bill users with varying levels of granularity based on quantum of computing power and storage used, and also the time of day. In that sense, it is probably more akin to the telecom system that exists around the world.
It is this computing grid that will finally make computing a utility. Today’s monikers like application service providers (ASPs) and software-as-a-service will dissolve into the more general-purpose commPuting-as-a-utility. This grid will provide computing and communications, and make possible the availability of the benefits of computers to the next billion of users, and simultaneously addresses the total cost of ownership issues for the first billion. This grid will shift the focus from managing hardware deployed across locations to encouraging the creation of innovative software applications much like HTML and the Web ushered in the golden age in content publishing in the mid-1990s.
Let’s begin by first understanding what grid computing is all about.
Tomorrow: Grid Computing