History repeats itself, first as tragedy, second as farce, according to Karl Marx. Is software as a service a tragi-comic re-run of the application service provider (ASP) model - a lovely idea that didn't work in practice?
The history of IT could be seen as one long class struggle. The workers - led by their departmental heads - struggle to get their hands on the means of production, to which IT limits access. Their latest weapon in this conflict - software as a (web) service - could allow them to bypass IT. But is it a force for good?
We have been here before. Back in the days of the mainframe, staff revolted against the computer managers, who became arrogant and territorial. The new SaaS model is much like the old mainframe computing set-up, where everything - data storage, processing, software provisioning - was handled centrally, and users accessed their computer services through a dumb terminal wired to the mainframe.
These days they can get any service they like through a PC (acting as a terminal) wired to the internet. Just as workers revolted against the mainframe when they found they could create their own Lans, they could now bypass the CIO once again - by buying SaaS.
But history has already repeated itself. This has distinct echoes of the 1990s, when ASPs promised much the same thing - but didn't deliver. There are two big questions, then. Will SaaS work better than ASP? And, far more importantly, how will this affect the political power of the CIO or IT director?
"SaaS is nothing new, and doesn't appeal. Neither do web services. It's the control issue that worries me," says one Socitm member, who didn't want to be named.
David Bradshaw, IDC's software and server research manager, damns SaaS with faint praise, saying there's nothing more nebulous than a new offering that's "in the cloud".
"This is the successor to the ASP market, and that really confused people. I'm not sure people want to return to the old ASP model. The ASP providers didn't have the technology to support their offering very well."
There seems to be a consensus among analysts that SaaS doesn't have to be like the ASP market. For a start, the infrastructure is more mature, says Bob Tarzey, services director at Quocirca. "It might be over-simplistic to suggest that SaaS is just the warmed-up corpse of ASP," he says.
"The bandwidth, security and datacentre facilities available to SaaS vendors provide a much better infrastructure, and there's better understanding and support built into the products and commercials by platform vendors."
Deployments are better too, he adds. The concept of virtualisation makes creating multiple tenancy more achievable. So SaaS is more likely to work than ASP, but could still be a disaster - for different reasons.
If SaaS became easy to buy, it could be seen by ambitious end-users as a way of bypassing the IT department. Worse still, if that strategy works, the board might realise that, as with ASP, SaaS could be regarded as a utility. And if IT is a utility, who needs a CIO or IT director?
Even strategic a applications, such as CRM, ERP and accounting software, could be ordered through SaaS, predicts Quocirca's Tarzey. That means trouble for the CIO if they don't work - and trouble if they do. In which case, it would be time for the CIO to fight a rearguard action, he says. "Departments can make independent SaaS choice, but bypassing IT completely may be a big mistake." Tarzey offers one form of defence: "Unauthorised applications can be blocked."
This reflects the new set-up for SaaS, says Bradshaw. "The IT infrastructure wasn't proven then, but it is more tangible now. ECS is more technically feasible because the bandwidth is there to support it, and the computing architecture is there. Licensing isn't an issue either."
Where the ASP market went wrong was that providers wanted to license the software, says Bradshaw. "These days the software industry is more behind the move. The software provider owns the software, and a service provider is often their channel partner." So a more workable model has evolved.
In times of economic flux and uncertainty, do people really want a slightly nebulous-sounding offering such as "in the cloud" SaaS? Yes, says Bradshaw. "It is definitely a better prospect in worsening economic times."
In-the-cloud SaaS (or software as a web service) could cut the cost of software significantly, admits Jos Creese, head of IT at Hampshire County Council. Given the constraints on public spending, surely that gives him more power, doesn't it?
But Creese is not convinced. "It's not business-critical given everything else going on," he says, "and the evidence of significant benefits and savings is weak. The offering is not compelling."
Besides, he says, there are better alternatives already. "Outsourcing does it for you anyway and if you insource, it creates complexities."
The other threat posed by software as a service is discontinuity. "Networks are still a potential weak link and a risk," says Creese.
So is this one of those great ideas that stays permanently at the concept stage? "I actually do think it will happen in due course," he adds.
But most Socitm managers Computer Weekly spoke to seem to think there are other priorities. The imperative in most organisations is not the delivery of software - that's pretty much in place already. What is more important is to reuse and redesign service processes. This, in turn, requires the IT department to think differently about the structure of services processes. "Service oriented architecture is a slightly higher priority and more practical, for now," says Creese.
As early adopters of SaaS know, it has one great advantage - it takes some of the responsibility away from the CIO or head of IT. On the other hand, it has one terrible side-effect - it takes some of the responsibility away from the CIO or head of IT.
It is the departmental computing/PC Lan revolution all over again, warns Carl Bate, who as CTO makes all the technology buying decisions for Capgemini.
"When individual business units are not catered for, they can go their own way," he says. "They can easily go the SaaS route." In a way, this can be good because it means software is paid for as an operational expense, he says, which can mean the money isn't coming from the CIO's budget.
A recent study by Gartner research vice-president Ben Pring showed that 75% of all SaaS is bought by business unit manages, rather than IT managers. Pring urges CIOs to get involved in the decision-making, pointing out: "It's happening in your organisation anyway, whether you like it or not."
Software as a service may be great because it is easy to deploy. But great for whom? The people who sell it find it a more profitable sale because one size fits all. End-users might enjoy a brief honeymoon period of easy deployment.
But this underplays the role of IT as a differentiator. If the board realises that software is a web service that can be ordered along with all the other office equipment, surely that undermines the CIO's position. Companies don't appoint an office equipment director, so why would they need an IT director?
The business pros (cost savings, flexibility) and cons (integration, lack of differentiation and a need for further investigation) are well plotted and argued elsewhere.
The politics of software as a web service do merit further discussion, and SOCITM president Richard Steel is probably best qualified to advise on this issue. "I think we need to be cautious and see SaaS in the context of all else that's radically changing," he says.
The blurring of distinction between web and other IT could have dangerous consequences, he warns, and CIOs must be wary that the perception of IT is changing with the way it is delivered.
"The web is an integral part of IT operations, not just an end-user service delivery channel," says Steel. "With access determined by identity and authorisation, rather than compartmentalised networks, that changes the approach to evaluation and benchmarking."
So SaaS is something you should keep a handle on, but not get too involved in, because there's not much credit to be earned. IT directors like Hampshire County Council's Creese would never say never to the idea of a software as a service - but not today, thank-you.
This was first published in August 2008