Clout of the cloud
Who will be the winners and the losers when cloud computing gains widespread takeup? E&T reviews the candidates.
Among the emerging technologies in enterprise IT, cloud computing enjoys the most commendation, and its proponents seem to have a positive reply for every critical reservation. 'Cloud computing' has become a buzz phrase, but there's plenty of evidence that it also represents a tangible platform for future applications, and offers the prospect of real operational and financial advantages.
As Frost and Sullivan research analyst Jennyfer Velez points out in her February 2009 report 'Cloud Computing: Not So Cloudy Anymore', definitions vary as to what cloud computing actually is. Velez herself defines it as a "flexible and scalable shared environment in which third-party suppliers use virtualisation technologies to create and distribute computing resources to customers on an as-needed basis, through the Internet browser".
What is becoming clearer as enterprises test cloud models is that there will be winners and losers in the cloud revolution. Among the winners will be corporate IT strategists looking for greater choice in the availability of applications they deploy, and also in the way in which those applications are provisioned. Some observers also feel that cloud promises to give the strategists greater flexibility in how they spend constrained budgets.
"Aside from technological considerations, cloud computing is really about choice," believes Reza Malekzadeh, senior director of products EMEA at virtualisation vendor VMware. "Typically, the majority of an IT budget goes on what you could classify as 'maintenance', and very little on what you could describe as 'innovation'. Organisations are devoting so much to keeping their critical systems up and running - both in terms of money and human expertise - that there is little left to fund improvements and upgrades. Cloud computing would change that scenario."
Malekzadeh compares cloud's ethos to the utility model we are used to for supplying our energy, water, and suchlike: he sees no reason why enterprises should not, in a similar fashion, procure at least some of their IT requirement from providers in the cloud; and that could mean multiple providers, each servicing a specific application: "In the same way that you can shift to a new utility that is offering a better deal, in the cloud scenario you can move to alternative IT suppliers".
If Malekzadeh is right, then budget formerly reserved for maintenance - keeping in-house systems working well - could be redirected. Applications that had been unaffordable might now be within reach, or it might now become possible to reconfigure in-house IT infrastructures to optimise performance. IT departments might even find themselves freed up to provide better user support, making users - especially those who need to work remotely - further winners in the cloud stakes.
But at the same time the necessity to keep as much bandwidth as possible available for cloud applications could result in IT managers imposing restrictions on employees' Internet access to non-work-related websites - especially those dispensing bandwidth-gobbling streamed media, such as YouTube.
"Most cloud computing applications are browser-enabled, and so bandwidth-hungry in both upstream and downstream. Their use involves constant data traffic, with information being downloaded, updated, and then returned to the cloud for storage," says Julien St John-Dennis, head of business products at ntl:Telewest Business. "Critical applications still relying on the UK's ageing legacy communications infrastructure could suffer downtime. As such, potential cost savings have to be balanced against a drop performance efficiency."
Bandwidth and cloud
The advent of cloud should also be welcomed by communications services providers because the cloud's browser-based dependence on WAN connectivity heightens the need for reliability and redundancy in online availability.
This is often cited as the 'Achilles' heel' in the cloud computing saga, because if the Internet link goes down, or runs slow because of access overload, then application performance is hit, and the business is hurt. Hundreds of users simultaneously working a cloud-based application would put extra strain on line contention that's probably already high.
However, enterprise networks built on conventional infrastructural models also have problems, and it is arguable that running critical applications on internally managed systems offers no better uptime guarantees than having them hosted externally. It could be further argued that, whereas internally hosted applications are periodically offline due to upgrades, modifications, and the need for server restarts, out on the virtualised platforms that cloud is based on, processing and storage capacity can be scaled up to meet demand at short notice, with no interruption of availability.
Back with the issue of bandwidth management, the cloud offers somewhat of a return to the tiered services model on which telcos based their datacommunications portfolios in the 1980s and 1990s: the more you pay, the more bandwidth and quality-of-service you get. "Service providers used to offer a broad range of wide-area communications options," VMware's Malekzadeh points out, "from dedicated leased-line links and virtual private networks, to packet-switched options like X.25. With the roll-out of broadband services these gradually became commoditised; now the previous market model seems to be coming round again."
Malekzadeh suggests that cloud reinvigorates this model, and presents communications service providers with the opportunity to launch tiered products tied to critical applications (SAP, say) that must have assured bandwidth allocation and management. Such a move would place them in a winning position in a cloud-covered ICT landscape.
One common misconception about cloud computing is that it's an all-or-nothing concept - that it only really makes sense if the IT function moves across to it en masse. Smart enterprises will realise that this is not true, and that while in theory a start-up could assign just about all of its IT to the cloud, more established enterprises will adjudge that it is only suitable for certain applications, and not for others. Email is a prime example of this, where regulatory compliances mandate that organisations retain full possession of their email volumes, and not outsource them to a cloud-located third-party. This requirement has implications yet to be tested for organisations that permit employees to customarily traffic corporate emails via Internet-based accounts with Google Mail, Microsoft Hotmail, and Yahoo.