Not long ago, a user on Quora asked industry experts, what happened to grid computing, noting that it seemed to be all over the news just five years ago but seems to have been overtaken by cloud computing. The user wondered If there was something different about clouds that made them better or if grid computing had simply died.
Grid luminary Ian Foster from the Computation Institute and Argonne National Laboratory jumped into the fray to address this question, noting as he has during our recent conversations that when it comes down to it, grid is about federation while cloud is more about outsourcing.
Foster contends that grid computing is thriving in scientific research, especially as there is a powerful need to distribute vast amounts of data and to enable on-demand access. He points to a number of long-running grid computing infrastructures such as Teragrid, stating that resources like these have been benefifcal in cancer, neuroscience, high energy physics, astronomy and beyond.
Foster goes on to write the following:
In industry, the term “grid computing” has been used, rather oddly, as sort of a synonym for parallel computing (e.g., Oracle 10-G) and sometimes to mean what the BOINC guys used to call (confusingly) “distributed computing”–i.e., harnessing idle desktops.
To use the electric power grid analogy, cloud is really about getting the supply side right, driven by new sources of demand and enabled by good distribution (broadband deployment). Scientific grids never do a really great job of supply: the economic incentives aren’t right. Thus I like to say that “cloud is about outsourcing; grid is about federation.”
As to whether cloud is just a renaming of grid: maybe both are a renaming of utility computing, as described by Doug Parkhill in his 1966 book, The Challenge of the Computer Utility.
For more on this, check out our video series featuring Dr. Foster and a number of others who spoke about the cloud/grid differences and what new developments are shaping both.