Google App Engine, Amazon Simple Storage Service and Apples MobileMe service have all experienced some high-profile service outages in the past few months. Is it safe to rely on the cloud?
Cloud needs to grow up
Cloud computing needs to move beyond its roots as a means for the Internet giants to sell some of their excess capacity and become a more mature, geographically redundant and robust computing infrastructure.
The popularisation of pay-per-use cloud computing is a recent by-product of the creation of web-scale applications by companies like Amazon and Google. Amazon Web Services, Elastic Compute Cloud (EC2) and Simple Storage Service (S3) were launched as ways of generating additional return from the companys huge investments in data processing and web services infrastructure required to support its core, low-margin retail business.
Salesforce.com and Googles App Engine were launched with similar motivations. Owners of the Internets heavy production infrastructure have realised that their assets can be, and need to be, sweated harder in an increasingly competitive market. The distribution channel that they have mastered for their core products gives them the opportunity to sell, as a by-product, some of the IT capacity that they use to run their businesses.
Notwithstanding Sun Microsystems Sun Grid Compute Utility, the major IT companies have been slower to see the opportunities in the cloud. They are, after all, making plenty of money selling shovels in the computing gold rush and providing outsourced IT services.
Recent announcements, however, confirm that cloud computing is indeed on the major IT vendors radars.
IBM is expanding its cloud computing capabilities by investing in data centres in the US and Japan adding to centres in Ireland, China and South Africa. IBM plans to employ 200 full-time researchers and invest more than $100 million in cloud computing over the next three years.
Hewlett-Packard, Intel and Yahoo are partnering to create new R&D centres for cloud computing. The companies will set up six Cloud Computing Test Beds to allow enterprises, universities and government entities to improve the software, hardware and management techniques needed to successfully operate cloud computing.
These announcements point to a recognition by the industry that there is a need for significant investment in boosting the reliability of cloud computing services. This R&D focus by the major IT vendors is warranted because cloud computing is hugely complex and prone to failure.
In this weakness lies its strength. The virtually limitless number of parameters for failure means that the only way to achieve high levels of reliability is to expect, and design for, failure at every point. The scale of the cloud creates both the necessity and the opportunity for redundancy on a massive scale which was the original purpose of the Internet itself, was it not? In theory, the cloud will in time become the most reliable computing infrastructure on the planet.
Sign up for Computerworld eNewsletters.