Cloud Computing on the Horizon

“I think there is a world market for maybe five computers.” Attributed to Thomas J. Watson Sr. (1874-1956) while he was president of International Business Machines, this apocryphal quote could turn out to be true in a manner of speaking if one were to rephrase the statement thus: ‘I think there is a world market for maybe five computer clouds.’ But would five clouds be enough? And if so, what services would these clouds offer and how might an organisation benefit from using them rather than more traditional on-premises solutions?

Cloud computing is an evolving model that is best described in terms of business and IT functionality. Today, most business and IT system functions are delivered or supported by on-premises IT systems. These systems are comprised of servers, networking equipment, and storage hardware, and are generally housed in a computer room, data centre, or other facility that is managed, controlled, or contractually bound to the organisation in question. If we replace the word ‘cloud’ with the word ‘Internet’, then things start to become easier to understand. ‘Cloud computing’ is, in effect, ‘Internet computing’, i.e. a computing model that uses Internet-based technologies to deliver a range of services.

It is, I believe, important to differentiate cloud computing from other computing models that are often perceived to be synonymous or of a similar nature. In conversations with Butler Group subscribers, the terms Software as a Service (SaaS – see Software as a Service Architectures, March 2008), ‘grid’, and ‘utility computing’ often crop up when organisations are thinking longer-term and strategically about IT provision, and now I am starting to be asked about cloud computing too. To my mind, these computing models are all part of the same emerging market, just viewed from a different perspective and each forming a different element of the ‘new IT supply chain’.

Although vendors tend to phrase things in their own terms, grid computing is generally considered to be the application of multiple computer systems to a specific scientific or technical problem that could not otherwise be addressed. When we talk about grid computing, there is a general assumption that all of the computing resource is co-located; but in practice, grid computing systems can be highly distributed, such as the SETI@home project – a scientific experiment that uses Internet-connected computers to Search for Extraterrestrial Intelligence (SETI) contained within radio telescope data.

While grid computing is suited to computing tasks that can be addressed through parallel computing techniques, the utility computing model takes the concept of metered public utilities – such as gas, electricity, water, telephone – and applies this to computing resources. Although for the customer there is no up-front cost to acquire IT infrastructure, organisations investigating this computing paradigm should be aware that they may have to pay a one-off ‘connection fee’, and maybe some kind of standing charge to cover a share of the fixed costs. Thereafter, the utility computing model follows the familiar ‘pay-for-what-you-use’ arrangement.

Those with long memories will recognise these computing models from the 1960s and 1970s, when time-sharing and ‘bureau’ services were common. Indeed, the early success of the Internet was partially due to the need to distribute computing tasks, and to share and optimise the use of resources. Today, these models continue to exist, albeit in a more open and affordable fashion. Not surprisingly, major hardware companies, such as CISCO, EMC, HP, IBM, and Sun Microsystems are all participating in the grid and utility computing markets at some level.

Cloud computing is evolving rapidly, as technologies mature and investments in computing infrastructure are put to work delivering new offerings and services. However, there has to be an economic factor or reason driving demand. Strange as it might seem to some, the current economic climate is probably the ideal catalyst for cloud computing, as organisations seek new ways to reduce cost while maintaining, or even improving, agility. Although organisations must be prudent and be able to manage risk well in the current climate, I believe that the cloud computing model does offer a glimpse of the way forward for the industry.

For the cloud to be generally useful it must, to some extent, be general purpose: it must provide a set of services that are broadly applicable and easy to consume. The most visible examples of the cloud computing model today are e-mail, ‘webtop’ productivity applications, hosting, and business application services. These have, by-and-large, been targeted at the consumer and small- and medium-sized businesses and can be grouped into the following categories: storage services, instantiation services, distribution services, security services, and application services.

A recent survey by Infosecurity Europe of 470 organisations revealed that 75% of them intend to reallocate or increase their budgets to secure cloud computing and SaaS over the coming year or so. However, in my own conversations and discussions with CIOs and senior IT managers, there is still strong concern relating to the availability and security aspects of SaaS models and service provided by the cloud. However, the constant challenge to do more with IT budgets is undoubtedly putting pressure on IT managers to try out cloud-based services, and so I expect we will see some growth in areas considered to be ‘quick-wins’, such as e-mail, collaboration services, and general business applications.

At the consumer-end of the market, Quality-of-Service (QoS), interoperability, governance, and service transfer all seem relatively unimportant at the moment, but in due course I fully expect regulation will have to play a part as users rely more heavily on the services offered by today’s providers. In the meantime, consumers and early business adopters should consider closely the Terms & Conditions of the services they use, and service providers should be transparent in the information they provide their customers.

In February, IBM revealed a series of products and services around its ‘Blue Cloud’ initiative. In addition to the cloud computing consulting services announced in 2008, IBM now offers a growing portfolio of services that promise better ease of use, economies of scale, and greater flexibility in sourcing and adapting to change. In a briefing to analysts, IBM talked about the technologies it sees as being central to cloud computing: virtualisation, Service Oriented Architecture (SOA), SaaS, and ‘request-driven provisioning’ – a term synonymous with the company’s On Demand initiative.

IBM’s cloud offerings are based around IT infrastructure, applications, and processes, but unlike other providers, IBM is championing a hybrid approach for the enterprise: ‘Private Cloud’ and ‘Public Cloud’. Private cloud – or ‘internal sourcing’ as the company describes it, is where the organisation owns and manages the IT infrastructure, applications, and processes. However, unlike many of today’s IT services, these will be delivered, managed, and maintained in a highly virtualised, highly standardised, and highly automated manner (using IBM’s products, solutions, and services one would assume).

With access defined and controlled by the client organisation, one can see how a private cloud can accommodate greater levels of customisation, increased levels of security, and to some extent, higher levels of availability (due to locality and business-determined levels of redundancy). If one then adds to this through the use of the public cloud – or ‘external sourcing’ as IBM describes it, then organisations get to conserve capital while exploiting economies of scale. Organisations are also likely to benefit from increased flexibility as a result of supplier choice (just like the utility model) and standardisation.

IBM’s foray into the cloud space with Amazon Web Services (AWS) announced earlier this year just goes to show how disruptive the cloud computing model can be; enabling dissimilar entities to coexist and partner in new ways. In this particular case, IBM is bringing a range of products to the Amazon Elastic Compute Cloud (EC2); thereby enabling organisations to utilise their existing licences within the cloud.

In the words of AWS: “EC2 is a Web service that provides resizable compute capacity in the cloud. It is designed to make Web-scale computing easier for developers.” To use EC2, developers create, upload, and then instantiate an Amazon Machine Image (AMI) containing applications, libraries, data, and associated configuration settings. Alternatively, pre-configured templated images – such as those being offered by IBM – can be used. In terms of billing, organisations only pay for the resources that are consumed, such as ‘instance-hours’ or data transfer.

Amazon and IBM are just two names from an ever growing list of companies developing service offerings for the cloud. Others including Google with AppEngine, and Microsoft with Azure, but my reason for highlighting Amazon and IBM is that it provides an interesting insight into the kinds of partnerships and relationships that may form over the coming months.

Organisations will, of course, have to restructure their operations, both in terms of business and IT, in order to benefit from the cloud computing model. However, the resulting streamlining, combined with the conservation of capital, will provide many businesses and institutions with significant new opportunities. Open standards, service oriented architecture, service management, scalable systems, and excellence in data centre operations are likely to govern the success of cloud computing in the longer term, but for now the world’s largest IT companies are placing their bets on cloud computing being the next ‘big thing’.