Over the course of the last 25 years I’ve watched several times as new technologies were introduced without regard for the investments already in place. Their fate was always predictable – failure.
As an example, consider ATM. Once hailed as the network of the future, ATM failed miserably in the market. It was complicated enough in it’s native form that simple implementations took months of planning. However, probably more importantly, it was a nightmare to implement in conjunction with an existing IP network. For this reason it never really had a chance.
The converse is also true, however. Technologies that fail outright or only succeed in a niche can flourish if brought to market in a manner that leverages existing investments. Apple’s iPod certainly wasn’t the first digital music player, but it was the first to come with a legal means of accessing existing music libraries. We all know what happened next.
GRIDtoday contributing author Greg Nawrocki recently wrote an article entitled Building the Perfect ‘Grid Sandwich’ in which he acknowledges that “The majority of grid implementations and even middleware solutions take the bottom-up approach, building the network and associated infrastructure and hoping applications are built that will utilize these services.” This is all too true, and as a result grid adoption has been slow to break beyond it’s HPC roots. Unfortunately, Greg continues in a predictably linear fashion “we need to concentrate on the top-down approach and actually build, or repurpose, applications that take advantage of potential Grid infrastructures.”
Building and “repurposing” applications is an immense investment. One that isn’t likely to spur adoption beyond those proven to have an acceptable ROI.
Of course, if you’re reading this you’re probably familiar with AppLogic and know that building or repurposing applications isn’t necesary anymore. Instead, what was required, and now exists, was a grid OS designed from the ground up to run existing applications. With this approach, Greg’s desire for a “Grid-enabled version of the Apache Web server” has already been fulfilled. In fact, AppLogic makes grid computing so transparent that many folks are already using Apache on grids without even knowing it.
To be clear, I’m not suggesting that there’s no place for software tailored to run on grids. However, that market is a niche and will always remain so.
I belive that for grid computing to achieve mass adtoption we need to leave the concept of “grid enabled” software behind and adopt the grid to the software people already use.