Cloud vs Grids vs Super-Computing

question_cloud1Clouds are gaining in popularity. The demand for data, analytics, and forecasting has grown significantly, and the future might belong to those who are able to predict it. However, to predict the future requires computing power – Lots of it. And cloud providers, hosting companies, startups, and big-technology companies are looking at providing this.

So what exactly is the cloud, and why will it provide the computing capabilities which has been dominated by super-computers over the last 2 decades. And why will cloud computing succeed where grid computing failed.

In May 17 1999, SETI@Home was release, and it gave the public a glimpse of how inter-connected computers could be leveraged to perform very large tasks. Grid technology was encapsulated in technologies like SunGrid and xGrid, but largely failed to gain traction. The internet was only starting to go mainstream, and computers were still expensive items.

A decade and a bit later in 2012, Cloud-computing is making headlines, and it seems that cloud-computing may succeed where grid computing failed. So what has changed since 1999?

  • Computers are cheaper
  • The Internet is much faster
  • VMWare and Virtualisation is making inroads into organisations
  • Hosting and Infrastructure companies are virtualising
  • Accessing virtual services like email, social media, SaaS is common place.
  • Increased awareness of online computing via Amazon Web Services, SalesCloud, Azure.

So will super-computing be replaced? Will there be reduced demand in running parallel jobs on multiple computing nodes? NO. There is significantly increased demand in running computer-intensive and parallel jobs. However the way in how a super-computer might be implemented will change. Instead of proprietary platforms, super-computer will evolve to open-platforms and be built on the cloud. The proprietary bits of super-computing will be the charging mechanisms for the utility.

Will grid computing be replaced? Grid computing will fade away. Grid computing addresses the same type of distributed super-computing that cloud computing would replace. The traditional super-computer might still serve a purpose for tightly-coupled applications which are difficult to distribute to the cloud or grid.

Consumers are not interested in a technology, but rather what they can do with it. In Cloud computing, this becomes more apparent with products like:

  • Database processing
  • Running an algorithm
  • Getting an answer