It was always inevitable that the adoption of cloud computing would fail to meet expectations, so feverish was the hype at one point. And lo, in 2010 cloud computing hardly turned the IT industry on its head as many had predicted it would.
In Information Age’s latest Effective IT survey, only 23.8% of respondents had adopted a utility computing service (as “infrastructure as a service” offerings were known when the strategy was first introduced to the survey) up slightly from 20.3% in 2009’s survey.
Still, the promise of cloud computing is no longer purely hypothetical. “There’s been a phenomenal amount of hype, but you also have realistic and reasonable cloud uses that are happening right now,” says Lydia Leong, an analyst at IT research group Gartner. “You’re starting to see production use, rather than just experimental use, and from mainstream companies, not just start-ups.”
Nevertheless, it is clear that not every organisation is sold on the proposition of consuming IT resources, on a utility basis, over the public web.
Above all, concerns about security and availability linger, and not without reason. Amazon Web Services, arguably the market leader, was struck by a partial outage in May 2010, after a car crashed into a pylon near one of its North American data centres. Other customers suffered intermittent access on another occasion due to problems with AWS’s electricity supply.
And it was not alone; various cloud and software-as-a-service providers including Microsoft, Terremark and NetSuite suffered outages throughout the year.
These concerns fuelled discussions of ‘private cloud’, in which a highly virtualised infrastructure allows organisations to manage internal IT resources as cloud services. Examples of large-scale deployments of these began to emerge in 2010, with defence contractor BAE Systems and investment platform Skandia being two notable cases.
But for both styles of cloud, real-world adoption has exposed some rather thorny systems management challenges. In fact, says Leong, many organisations are finding that there is more systems management involved than they expected.
“Enterprises hope that cloud will be ‘magic’ infrastructure,” explains Leong. “They think they’re not just offloading the infrastructure, but also a portion of the management that is normally associated with having to run [internal] infrastructure.” That is not the case, she says.
In IT, a user’s headache is a supplier’s opportunity, and mainstream IT vendors began to buy into the nascent ‘cloud management’ space in 2010. In November, for example, computer-maker Dell purchased Boomi, a start-up that helps enterprises quickly integrate on- and off-premise applications. Similarly, in January this year, managed services vendor Rackspace acquired CloudKick, a San Francisco-based venture that allows businesses to manage internal and external virtual servers through one dashboard.
Leong expects this trend of infrastructure heavyweights buying into the cloud management space to continue in 2011.
Another lesson being learned by early adopters is that the on-demand pricing model that is commonly associated with cloud computing, and is often said to be one of its key benefits, is not necessarily economical.
According to Leong, a high proportion of cloud customers are simply opting for fixed-term contracts. She says that for many organisations demand is in fact rather predictable, and that the discounts offered by suppliers make signing up long term more cost-effective.
Cloud computing does not end at so-called infrastructure-as-a-service offerings, and in 2010 various suppliers experimented with their own interpretations of the term.
Having in the past denounced cloud computing as “water vapour”, in 2010 Oracle CEO Larry Ellison appeared to embrace the concept. In September, his company launched its ExaLogic system, which it described as an “elastic private cloud” stack containing 30 servers and assorted middleware. It is powered by hardware that Oracle acquired through its purchase of Sun Microsystems.
Critics dismissed this description as “cloud washing”, however, arguing that ExaLogic is no more of a private cloud than a mainframe is.
Meanwhile, there was some movement in the space between software as a service and IaaS. ‘Platform as service’, as it is known, allows software developers to build their own applications using functional components and hosting infrastructure from a third party.
In February 2010, software baron Microsoft entered the fray with Azure, a cloud-hosted platform where developers can program and host applications using the popular .NET framework.
Salesforce.com launched its PaaS offering, Force.com, back in 2007, but while it attracted some high-profile users including Japan Post, adoption did not match the success of its CRM product.
One problem was that Force.com employed its very own programming language, called Apex. Although this was similar enough to Java for programmers to learn easily, it meant that existing code could not be ported across to Force.com.
To remedy this, Salesforce.com turned in February to virtualisation market leader VMware. Together they launched VMforce, which combines the Force.com platform with the Spring software development framework, the company behind which VMware acquired in 2009. This allows developers to move Java applications they have built using Spring on to the Salesforce.com platform.
Despite the fact that VMforce was announced in April, it remains unclear whether it has attracted the support of software developers, an essential victory if it is to be used in earnest.
In December 2010, Force.com also added support for Ruby, a language that has grown in popularity with developers in recent years, through the acquisition of development platform Heroku.
But while suppliers are still coming up with new permutations of the cloud model, the previously interminable discussions on the meaning of cloud computing seem to have been laid to rest. Whether this is because the industry has reached a consensus on the meaning of the term or whether it has simply tired of the semantic wranglings is not entirely clear.