One of the Yale researchers has warned that cloud-based systems might melt down with the systems becoming more and more complex.
Bryan Ford has written a paper, which he is going to present to the USENIX HotCloud 2012 conference soon. The paper says that with the use of cloud computing now becoming more mainstream, major operational “meltdowns” might arise. The matter is that everything will get quite complex, and complexity will cause an accident.
Ford explained that as diverse cloud services share more fluidly and aggressively multiplexed hardware resource pools, the probability arises that unexpected things will happen, including unpredictable interactions between load-balancing and other reactive mechanisms. This may result in dynamic instabilities, also known as “meltdowns”.
According to the experts report, it was a little like the intertwining, complex relationships and structures which could promote global financial crisis. He pointed out that new cloud services may emerge, which actually resell, trade, or speculate on complex “’derivatives” like financial trading industries.
Such components will be maintained and deployed by different companies, which, due competition, won’t share details (if possible) about the internal operation of its services. As a result, the cloud industry might face speculative bubbles. The experts predict occasional large-scale failures due to composite cloud services which have weaknesses that do not reveal until those bubbles burst.
Meanwhile, there’s no solution to the problem. The only advice that the experts can give is that providers should release detailed data about their system dependencies to some special 3rd party that offers cloud reliability analysis services.