March 29, 2013 | Written by: Charlie Liang
Share this post:
As someone who holds a master’s degree in atmospheric science, I never imagined that the word “cloud” could one day be used ubiquitously in the world of IT.
I always thought we would associate the word with the white, fluffy cotton-like shapes or dark, ominous thunderstorms we see in the sky. Whoever coined the term “cloud computing” was probably trying to be creative (maybe he or she also had an atmospheric science background?). From a certain perspective, the term is rather eye-catching. But in general, I think it has created a lot of confusion among non-IT folks. Quite a few people seem to think “cloud computing” is somehow related to our planet’s weather.
According to a article in Business Insider, a survey conducted by Wakefield Research in August of 2012 including more than 1000 American adults showed that “51 percent of respondents, including a majority of Millennials, believe stormy weather can interfere with cloud computing.” (Click here for the original article.)
Well, to set the record straight, daily weather changes do not in any way affect cloud computing. But still, I’ve been thinking a lot on how cloud computing will affect the way we do weather forecasting.
The first thought that came to mind was whether the growing prevalence of cloud computing would create a drastic change in the way weather forecasting agencies (private or government-owned) invest in their infrastructure.
I think sooner or later, policy-makers will start to discuss (or debate) the necessity of spending millions of dollars investing in the high-performance computing infrastructure that is needed to run the numerical weather forecast models. If we can instead rent the whole thing from a cloud service provider at a reasonable price, there’s really no point in investing millions of dollars every few years in upgrading the relevant infrastructure.
Nevertheless, as mentioned earlier, this all hinges on the price tag that the cloud service providers are willing to offer, which at present would likely alienate most organizations. But as more companies jump on the bandwagon, market competition will become increasingly intense, and price cuts will definitely occur.
In the not too distant future, I believe it will drop to a point where it will be more advantageous “to rent” than to “buy and maintain” the powerful computers.
Just how much does such a supercomputer cost? Well, to give everyone a clearer perspective, a few months ago, Taiwan’s Central Weather Bureau allocated a staggering $16.7 million to buy the PrimeHPC FX10 supercomputer from Fujitsu.
In the world of weather forecasting, I think the advent of cloud computing will be a win-win situation. Cloud service providers will generate a lot of revenue from this new rental business, while weather forecasting agencies can utilize the savings to conduct further research, such as developing more advanced computer models.
So although the weather doesn’t affect cloud computing, I think it’s safe to say that the latter will definitely help us to understand the former in a more cost-effective way.