In the past, grid computing has been used primarily in high-performance computational computing. Is it applicable to mainstream businesses today?
It's applicable. Most companies today seem to be struggling with the notion of wanting to grow their infrastructure without just buying hardware. This is driven by budgetary and management constraints; they just have too much stuff out there already. The promise of grid technology is that it will help companies consolidate and virtualize environments so that they'll be able to do more with less. What businesses are adopting grid computing today?
Financial services companies are adopting grid technologies now. In a sense, they fit the model because they're doing computational computing. In financial services companies, people want to do a calculation and get a result very quickly. The difference is that, in the past, grid computing was focused on large-scale computing, batch-driven workloads. The financial services workload is more transactional in nature. The turnaround times need to be fairly rapid. Also, it's a more service-based infrastructure, where the user is requesting a service to provide a result of a calculation. That's different from running an analysis and then having to look at output files and so on afterwards.
There seems to be a lot more interest in grid computing, as businesses are looking to Web services to consolidate these kinds of analyses in their companies. They want to consolidate them
At high levels in a company, executives and their IT teams need to be cognizant of what they can do with grids today. They should keep an eye on what other businesses are intending do with grids tomorrow. Today, a lot of grid technology is still focused on techno-computing, such as doing assimilations or computational workloads. I can see a movement toward general business applications being run in these virtualized environments. What other types of companies are a good fit for grid computing?
It's mostly application-driven. Internally, a company has to decide which applications would provide business benefits when scaled up. That will help to direct them as to where they should invest in their infrastructure to support the application. If they see that they can get a good ROI from running more simulations or more of a particular computational activity at a remote site, they might see that they need to invest in software or services to build the software infrastructure. Or they might see that they need to invest in networking infrastructure to help support the data transfer.
If you can identify the application that can impact the bottom line the most and work on that one first, you can incrementally build the infrastructure needed to support grid computing there. It's an approach to building infrastructures, but you still have to build that infrastructure.
Does a business always have to build? Can existing infrastructures support grid computing?
They must ask themselves: Can our infrastructure support that application in a remote-computing model? Does it have the ability to move data around, have data availability for the application and have the environment available to execute the application? If I have that today within my organization or have applications that can easily be enabled to do that, then I'm ready for grid computing today. If the costs associated with making that happen are too prohibitive, perhaps grid computing isn't the right approach today.
It's going to become easier to implement. Some of the standards around data integration, managing the streaming and caching of data will become more mature. Then we'll see more technologies that will help companies distribute data around the enterprise in a way that creates a good return on investment.
FOR MORE INFORMATION: