Back in late June, Amazon.com’scloud service experienced an embarrassing outage after a series of thunderstorms knocked out power near its East Coast data center, GigaOM reported.
Services from Instagram, Quora, Heroku, Pinterest, Hootsuite, and Netflix were all affected, and a dating service called Whatsyourprice.com quit Amazon over the issue, as Information Week reported.
Clouds aren’t supposed to do that.
Even when data centers go out, cloud customers are told to copy or mirror their services in multiple data centers. Just as this file is copied or cached thousands of times across the Internet, so are cloud services expected to be replicated over time, becoming more Internet-like.
That’s the goal, anyway. Cloud computing (explain this)isn't supposed to be a data center, but a network of data centers, a global network like the Internet that will offer redundancy, multiple paths between people and resources, the security of knowing you can never be knocked offline, as Whatsyourprice was.
We are told that what stands in the way of that goal are closed systems and closed standards. VMware’sESX hypervisor, the market leader for virtualization, letting programs written for different operating systems run on shared hardware, is proprietary. Amazon’s Application Program Interface (API), considered a de facto industry standard for public clouds whose use is rented, is proprietary. So is the Microsoft Azure cloud operating system.
Against these market leaders stand a growing ecosystem of open source alternatives like the KVM hypervisor, the OpenStack infrastructure initially sponsored by Rackspace Hosting and now supported by Red Hat (among others), and Joyent’s SmartOS, based on Sun Solaris.
Open source advocates want to build entire cloud “stacks” on open source. A cloud stack would include an operating system, a hypervisor, infrastructure, and a platform with computer languages and middleware.
They argue this will enable the kind of interoperability found on the Internet, which is also based on open standards.
But even here users have choices. Do you use SmartOS or Red Hat Enterprise Linux? The KVM hypervisor or Xen? OpenStack or CloudStack, which is offered by Citrix under the Apache license, as described in this press release?
All these questions are still being thrashed out, and all these software products are at different levels of maturity, with projects seeking the attention of developers to build out the clouds of their dreams — meaning customers easily get confused.
Rather than stay confused, many customers look to trust one company’s vision, whether it’s that of VMware, the virtualization market leader, Amazon, the public cloud leader, Google, which pioneered many cloud concepts and has now begun offering its own Google Cloud, orInternational Business Machines, which calls its offering SmartCloud.
The problem is that when you trust one company you wind up back at square one, locked into a proprietary technology that may not grow into the global Internet cloud you want because you're depending on one company to develop it.
I call this the cloud runaround. It’s the big technology debate of our time. It’s what will define this decade as devices defined the 2000s, the Internet defined the 1990s, client-server did the 1980s, and personal computers did the 1970s.
The winners will gain untold riches. The losers will get jobs working for the winners. The outcome is unknown. Everyone’s crystal ball is murky.
The fun has just gotten started for technologists and investors alike.
—By TheStreet.com Contributor Dana Blankenhorn
CNBC Data Pages:
- Dow 30 Stocks—In Real Time
- Oil, Gold, Natural Gas Prices Now
- Where’s the US Dollar Today?
- Track Treasury Prices Here
At the time of publication, Dana Blankenhorn was long Microsoft, Google, and IBM. TheStreet’s editorial policy prohibits staff editors, reporters, and analysts from holding positions in any individual stocks.