INSIGHT: Why cloud computing will NOT wipe out enterprise data centres
- 12 June, 2015 06:02
I still remember the time when the first CD players came out in the mid 80s. By the mid 90s vinyl had virtually disappeared from the shelves.
Same with the Flatscreen TVs. In this case it went a couple of years faster until the CRTs had been virtually wiped out around 2010-2012 - "Nothing is as powerful as an idea whose time has come."
Although I am apparently a big believer in Computing Clouds, I do not observe a similar wipe out of traditional enterprise IT and data centres (yet).
When I look around at technology in general, the trend seems to be even the opposite: decentralisation instead of centralisation.
• 3D printers are said to liberate people and protect people’s sovereignty.
• The Tesla power wall announced a couple of weeks ago aims to eventually make people’s homes independent from any national power grid.
• Bitcoins …
Pretty stiff arguments against centralisation.
So where are computing clouds? What could be wrong with them? I believe they oversaw the data.
In the year 2011, McCrory described an a meanwhile famous Blog Post the qualitative characteristics of data that he called ”Data Gravity”.
”Data Gravity” is a metaphor describing the economics of data, demanding data to better stay where it is and to not to ship it around, no matter how big or small the amount of data may be.
A finding that is supported by Jim Gray who stated that compared to the cost of moving data, everything else would be negligible (in D. Patterson, “A conversation with jim gray,” ACM Queue, vol. 1, no. 4, pp. 53–56, 2003).
As consequence, McCrory states that data must have something that is comparable to a gravitational pull that pulls services and applications to it rather than the other way around: ”Data Gravity”.
This blends with the Map Reduce programming model where computation, for example batch jobs written in Java or Python, are brought close to the data rather than the other way around.
Under the assumption that ”Data Gravity” exists it seems reasonable to question whether computing clouds, that centralise and rationalise resources for computation, that are seemingly mobile, are the landmark innovation that is required to wipe out inefficient onsite data centres.
This is further amplified by the ever increasing number of powerful heterogenous mobile devices.
Disregarding data gravity, there are of course many more issues around data, such as data residency, privacy, confidentiality and so on…
The alternatives to computing clouds, onsite data centres, do not seem to go away as quickly as everyone had hoped three years ago. Recently also Microsoft chosen to ‘renew’ their ‘strong’ commitment to Sharepoint Server. For now.
By Joerg Fritsch - Research Director, Gartner