High performance computing (HPC) has been dominated by the medical research and universities across the globe. As the market size grows and is expected to reach $15.6 billion by 2012, Joseph Kremer, president, Public for Dell Asia Pacific Japan region, believes that the market is much bigger than it appears with different sectors investing in HPC services. ?I think we are quite excited about the growth opportunities that the Indian market represents. In the last year, our public business and overall Indian business has grown,? he says. In an interaction with Diksha Dutta, Kremer discusses the company?s key focus areas and challenges in the Indian market. Excerpts:

What is your customer base in HPC and how have they benefited from your services?

The single biggest example is the $10 billion Atlas project globally. It will be the largest HPC computing series of clusters in the world?and Dell is a primary partner. It is organised primarily around universities all over the world. In India, one of our customers is National Institute of Technology, Calicut. Last year, we bought Perot systems, a services business?it added more than 30,000 co-workers and made us the largest hospital outsourcer globally. Dell-Perot signed a large IT outsourcing deal with one of the largest hospital chains and healthcare providers (Max Healthcare) in the country. So their customers are in our segment. We work with a lot of PSUs?whether defence PSUs or oil sector. This is both in case of high performance computing or any other security solutions. Some of the premier engineering institutes in India are users of Dell.

But where is the potential market?public sector or private sector?

It?s a mixed bag. Probably the largest community would be universities all over the world, but they are by no means alone. It includes a number of government organisations that have research, along with oil companies and commercial organisations. There are also stock trading companies who might have smaller clusters but they do HPC. It includes people who do outcome based trading and research around stock pricing etc. So it is broader than you might think.

The market is quite huge but as a community there are more research organisations and universities into it. In India, Defense Research and Development Organisation (DRDO) labs, Council of Scientific and Industrial Research (CSIR) labs, engineering sector, manufacturing in design?all use HPC. That is really where the buck of the market is. According to IDC, the current HPC market is worth $10 billion, with IDC predicting growth to $15.6 billion by 2012. This is a fairly substantial market. But, what I am observing is that medical research is one of the biggest sectors in HPC.

What are the challenges you face in the Indian market?

Considering not just India but Asia, the researchers are experts in their field and are often also experts in writing software to process the research. However, it is less often that they are experts in IT. The challenge with them is that they might need people like Dell to schedule jobs as we have this big supercomputer sitting here and often a whole host of researches are serving it and they are all submitting jobs constantly. And if you are not thoughtful about the software you used to manage those jobs, it can either be inefficient or actually stop working. Thus, managing that can be a challenge.

Another important challenge is managing storage. You can take an HPC set up and reduce its effectiveness by 50 or 60% because of poor design storage. Sometimes the only thing we do for customers here is that we just change their storage design and they get a huge leap. These two areas are common in Asia.

How do you face competition against Microsoft, Oracle, IBM and HP?

I probably would not include Microsoft as a competitor in HPC. They would potentially be more of a partner. The other competitors you mentioned is fair enough. I think the main difference with us is that we have restructured our business and we have an entire team whose only job is to focus on universities and HPC.

We have an entire salesforce dedicated to HPC. We tell our customers to buy the capacity that they need, instead of buying the whole proprietary solution. For example, a research might start out with 50 servers and appropriate storage. After 6 months, they could add 10 more servers as our model is a scalable and linear model.

A lot of researches have big systems with only 50% usage, incurring high cost of maintenance and software. But our focus is on smaller packages which are industry standard?you can accelerate the growth.

How do you add value to your customers with HPC services?

Our focus is to help research institutions improve their research outcomes through better use of technology. In areas of managing the storage, you can imagine the massive amount of data that has to be accessed for a lot of this research. If it is not an efficient storage infrastructure, it can slow down the research. Thus, the mission for us is that with a limited research budget, how do you help and make the research advance more quickly. We concentrate on data management, efficient infrastructure and the design of an IT department in an HPC environment.

We have introduced a series of new servers and storage devices and there is a reason. Last year, there was more data created than the storage. There was 18 exabytes of data created and 90% of it was unstructured. Further 90% of that will never be accessed together. So you can think about the cost of storage data.

The processing power of the new servers is more than 200% faster than the ones they have replaced. So if you are running a large IT firm, you can spend less money on more capacity.