Forget ‘cloud-first’: Australia’s public sector is following the globe into ‘cloud-only’


opinion/analysis by Renai LeMay
31 January 2014
Image: gdsteam, Creative Commons

Burned by a decade’s worth of failures of major on-premises IT projects, 2013 was the year that Australia’s State Governments almost universally declared they would take a “cloud-first” approach to IT procurement. But there are already signs that the next stage of this process is underway, and that “cloud-first” may inevitably become “cloud-only”.

Francis Maude (pictured) is not a politician known for holding his tongue or being overly diplomatic when he thinks he needs to make his views plain. As the son of a former British cabinet minister and editor of the Sydney Morning Herald (Maude spent time in Sydney as a child), with a history as a criminal lawyer and now thirty years in the British Parliament, Maude has weathered his fair share of controversies. He has emerged from that process with a habit of dealing with issues forthrightly and somewhat forcefully. He typically says what he means.

In a major speech given to the Sprint 14 conference held in London this week, Maude, who since 2010 has held the post of Minister for the Cabinet Office, didn’t disappoint. One of the politician’s key messages was an extremely blunt one, with a high degree of importance not only for the UK’s IT industry, but also global technology vendors in general. Put in plain English, Maude has had a gutfull of massive government IT projects failing continuously, and he’s determined to do something about it.

The MP said that previously, the UK government spent more on IT than any other country in Europe except Switzerland. “Although I think that included the cost of CERN,” the Minister added. “They were looking for the God Particle – but over here, we were left with an ungodly mess.”

In the old world, Maude said, the country was procuring applications before they had been designed – or over such a long period of time that the technology was out of date before it was delivered. “To re-visit the film metaphor: we were promised ‘It’s a Wonderful Life’, charged a ‘Fist Full of Dollars’ – and then a ‘Few Dollars More’ – but we were left with ‘Titanic’,” he told the conference. “For too long, big IT and big failures have stalked government. Now we want to see a new world, a start-up world, where what you can do matters most and where value includes both cost and quality.”

To address this issue, Maude revealed that the UK Government has mandated a list of what he described as “IT red lines”, which he said he would be “unashamedly militant about enforcing”.

The new rules state that no IT contracts will be allowed to exceed £100 million (AU$190 million) without a powerful reason; hosting contracts will not last for more than 2 years (“the cost of hosting halves every 18 months, why commit to a longer contract?”); there will be no automatic contract extensions without a compelling case; and companies with a contract for service provision will not be allowed to provide system integration in the same part of government (“there is a conflict of interest here, and contracts are too opaque”). Maude said:

“The whole point is for Whitehall to look beyond the oligopoly IT suppliers – the legacy technology giants. We want the right technology at the right price for taxpayers – whether that’s from an innovative big supplier which gets the new ways of working with us, or a start-up.”

Now, to politicians and the general public, these fighting words seem entirely logical. After all, what IT project or service could possibly be worth around AU$200 million to procure, in the age of the iPhone, Gmail and Amazon Web Services? However, to major IT vendors such IBM, Oracle, SAP, HP and the public servants who have continually purchased products from them over the past several decades — in first-world countries such as Australia, the US, Canada and others — Maude’s comments represent something of a declaration of war, and a clear signpost that governments globally, and perhaps shortly corporates, are now shifting towards IT procurement models which focus on cloud computing only, to the exclusion of traditional frameworks.

If you examine the recent history of major IT projects within the Australian Federal Government and all of our State Governments, what you will find is that almost universally, they have been based along the lines that Maude severely criticised this week.

Australia’s Federal Government has had a slew of recent major projects which have cost hundreds of millions of dollars. The ATO’s Change Program, built on Oracle software by Accenture, cost $800 million. Immigration’s Systems for People program, built on Oracle software by IBM, cost around $670 million. Defence has several initiatives in the works which are likely to cost several hundred million dollars, including its long-running revamp of its ailing payroll systems. Customs’ Integrated Cargo System was of a similar size.

Similar projects can also be found throughout Australia’s State Governments. Queensland Health had its payroll systems upgrade and its statewide e-mail and collaboration platform. Victoria has the Link project being carried out at Victoria Police, the myki public transport smartcard effort, the education department’s Ultranet initiative and the HealthSMART health systems rejuvenation. Most of the states have shared services programs around payroll and financial management.

And what we find almost universally in these projects and in Australia is that they have not been delivered as expected.

Most of the major Federal Government IT projects mentioned have suffered major budget and delivery timeframe overruns, and in some cases have gone off the rails so badly that they resulted in major service delivery failures, such as when the ICS failure caused pile-ups of cargo on wharves a decade ago, or Tax’s Change Program revamp played havoc with the agency’s ability to conduct its core tax return business.

In November 2011, Victoria’s Ombudsman handed down one of the most damning assessments of public sector IT project governance in Australia’s history, noting total cost over-runs of $1.44 billion, extensive delays and a general failure to actually deliver on stated aims in 10 major IT projects carried out by the state over the past half-decade. A similar ICT audit has been delivered in Queensland, and that state, NSW, Western Australia and Victoria have been forced to substantially modify or even cancel their whole of government IT shared services initiatives, due to their continual failure. Even the Northern Territory has its problems.

It is precisely these kinds of failures which has gotten Maude hot under the collar. The UK Government has, after all, suffered its own string of major IT project disasters, with the National Health Service’s National Programme for IT, with its £2.2 billion current cost (mainly paid to CSC) headlining the list.

UK watchdog the Major Projects Authority published an extensive list in May 2013 of UK Government IT projects which were at risk. The list contains quite a few items which represent very familiar stories to Australian technologists, including the Ministry of Justice’s IT Shared Services program (ERP) and the Ministry of Defence’s core infrastructure replacement project.

What Maude is attempting to do with his new ‘red lines’ strategy is to avoid the UK Government getting into these kinds of projects in the first place, and re-focusing is towards smaller projects that can actually be delivered with substantially less risk.

The striking thing is that this is precisely what is happening in Australia as well.

In March last year, for example, the South Australian State Government issued a new whitepaper designed to provoke discussion of its future ICT strategy, promising as part of the document that from then on, it wouldn’t pursue “big ICT projects” any more, with all technology-related initiatives to last 90 days at most.

Similarly, State Governments such as Queensland, New South Wales and Victoria, which have both been pummelled by the failures of massive IT projects, losing hundreds of millions of dollars on individual projects, have mandated “cloud-first” approaches to IT procurement which should result in massively smaller IT contract costs, as IT is procured on a service basis, avoiding large monolithic software customisation projects and instead customising their business processes to meet the demands of cloud computing platforms.

The difference between such moves in Australia and Maude’s comments in the UK is only one of tone, driven by necessity. Because the UK does not have the same large State Governments that are responsible for many core government functions in Australia, the UK Government has tended to have IT projects which are larger than those in Australia. Hence, the amount of money which has been lost in problematic major IT projects is larger.

But the trend is clear: Maude has mandated that no new IT projects above AU$190 million will be initiated, unless there is a “powerful” reason for doing so. In Australia, most State Governments have stated that they will focus on the “cloud-first” deployment model and on smaller IT projects. These two different philosophies are clearly steps on the same journey; the UK is just a little further down the track, having lost patience with the idea of major IT projects earlier.

If you examine the likely impact of this trend towards smaller IT projects, what is clear is that it will universally drive departments and agencies away from on-premises or even managed services deployments of enterprise software and towards the cloud computing model. Government managers will shortly stop asking whether a project can be delivered through a cloud computing model and start asking why it can’t. There will be no other way for UK departments and agencies to come in under Maude’s £100 million cap. And history has shown that it is almost always much easier for large organisations to customise their business processes to meet the demands of software platforms, rather than to customise software to meet the demands of their business processes. This is the lesson the UK and Australian Governments have learnt painstakingly over the past decade.

Steven Sinofsky, the former President of the Windows division at Microsoft, wrote on his influential blog Learning by Shipping in December last year that a “cloud-only” model was likely to become the norm in 2014:

“Enterprise software in 2013 was a dialog about on-premises or cloud. In 2014, the call for on-premises will rapidly shift to a footnote in the evolution of cloud. The capabilities of cloud-based services will have grown to such a degree, particularly in terms of collaboration and sharing, that they will dwarf anything that can be done within the confines of a single enterprise. Enterprises will look at the exponential growth in scale of multi-tenant systems and see these as assets that cannot be duplicated by even the largest enterprises. Design: Don’t distract with attempting to architect or committing to on-premises.”

Steve Hodgkinson, the director of Ovum’s Government practice in Australia and New Zealand, did much to support this vision, when he wrote on his blog this month about the kind of whole of government IT strategies which have bedevilled Australia’s states:

“IT strategy in government too often resembles a horror story because of a passion for big projects which become mad experiments that go bad. Under pressure to cut costs, whole-of-government CIOs pursued adventurous and experimental shared services and common application and infrastructure strategies. While the business cases may have looked compelling, the results were often disappointing.”

“… The real value of cloud services for government is to provide an alternative to such experimental whole-of-government IT projects. A cloud service can be deployed quickly in an individual agency to prove its relevance and usefulness. Once a service is operational and the agency is confident that it performs, then it can be adopted by other agencies on an opt-in basis. While the initial implementation may be experimental, it is at least relatively quick and inexpensive. The subsequent adoption of an operational cloud service by other agencies is an incremental, low-risk process compared to mandated whole-of-government shared service projects. Agencies simply choose to become a consumer of a cloud service that is already known to work.”

“… Whole-of-government CIOs should abandon big bad consolidation, standardization, and rationalization experiments. The focus for 2014 should be on the promotion of cloud services adoption and intra-agency collaboration via knowledge sharing platforms and pragmatic approaches to standards and information interoperability.”

What we’re seeing here is the fact that analysts, commentators and ministers are baldly stating, in public, that the era of big bad government IT projects which failed is over, and that the next phase of government IT should specifically focus around cloud computing projects instead, with their significantly lower capital expenditure requirements and lesser risk.

In jurisdictions such as the UK and South Australia, which have explicitly signalled plans to outlaw major new IT projects, what this signals is a shift towards cloud computing in all instances of new projects. With the limits that Maude is discussing in the UK, and the extreme reluctance on the part of Australia’s emerging cadre of state IT Ministers to approve major new IT projects due to the failures of the past, use of cloud computing platforms — software as a service, infrastructure as a service, platform as a service, storage as a service and so on — will emerge as the only way to progress necessary IT projects, rather than just an optional way.

We’re already seeing this approach to a certain extent in Australia. In New South Wales and Queensland, the states’ IT Minister have virtually mandated that new pan-department email platforms be based on cloud computing environments rather than in-house deployments, due to failed in-house models in the past. It won’t be long before that approach starts to percolate throughout other areas as well.

In another example, Ministers all around Australia will have been eyeing the success which the NSW State Government has claimed in its high-profile deployment of a cloud-based ERP consolidation project at the NSW agency of Trade and Investment. The State has claimed that so far the project has been delivered “on time and on budget”, but with a large chunk of the work still to go. The whole project is only worth around $15 million; a pittance compared with the hundreds of millions spent on comparable on-premises projects in the past. And the time to deployment has been extremely rapid.

I don’t think it will be long before Ministers of all stripes in Australia start refusing to approve major IT projects unless it can be demonstrated that they can be delivered relatively easily as a service rather than as a monolithic in-house install. Due to its larger scale and the extremely complex nature of major IT projects in its major agencies, the Federal Government will be a long-term holdout to this approach. Due to the significantly degraded level of their IT infrastructure, State Governments will be first into the water, while cloud chips away around the edges in Canberra. But it will happen. Eventually, if IT projects can’t be done quickly and nimbly, they won’t be done at all.

Governments are much more transparent than corporations when it comes to disclosing IT infrastructure; much more is known about major IT project headaches in the public sector than the private. But I suspect the private sector will shortly find itself following the public down the “cloud only” path as well. Examples of botched ERP deployments in major corporations in Australia are not hard to find, after all; showing that many of the same problems exist across the public and private sectors.

In the long-term, it’s hard to know what the implications of this shift will be. The last time Australian Governments underwent such a dramatic change in IT project and service delivery was in the late 1990’s and early 2000’s, when a wave of comprehensive IT outsourcing contracts were signed with major suppliers such as IBM and EDS. Since that time, those contracts have been broken up into smaller parts and a raft of new technologies have made their way into IT departments.

But what we’re seeing here with “cloud first” is the same kind of massive wave of change. I don’t think anyone can predict what its overall impact will be in five to ten years. But one thing is clear: Things probably couldn’t get worse. If you doubt that statement, I encourage you to examine Queensland Health’s payroll systems upgrade project and ask yourself whether it wouldn’t have been delivered more successfully (or successfully at all) from the start if the Queensland Government had banned big vendors from day one and opted for a more nimble, more standardised and lower cost cloud computing model. The answer, one has to surmise, is clearly “yes”.


  1. I see this as a good sign. There’s no justification in 2014 for hundreds of millions of dollars (let alone billions – I’m looking at you WA State Government) for what seems to commonly boil down to front end applications with backend databases.

    The problem seems to be that big IT has not moved from the old days of large and complex software projects developed by the few technical experts. Not only does the old model produce unwieldy behemoths that require armies of IT support personnel but in a time of increasing speed of change, it almost guarantees legacy is built into new projects from which extraction, not surprisingly, will cost another $100m :)

    What Francis Maude may be alluding to is the need for lightweight, low-cost applications, databases and reports that take advantage of new and emerging technologies. Requirements change, customers demands change, opportunities present themselves. Old school solutions practically and financially preclude change. And there is a certain amount of ‘lock-in’ with a big IT provider to ensure future revenue. Sell once – collect money multiple times? Seems like big IT would be worried about that more than anything.

    We used to have a general guideline around our IT projects. The specifications, project intent, requirements and functions of any IT system seem grow stale in 12 months and need to be refreshed or completely redone every 18 months. This can’t be done with high-cost, old school solutions.

    • I agree that dumb decisions are being made over and over because its public money right? Actually corporates can be stupid too…

      But cloud isn’t a single miracle pill – its a little more complicated than that. Public data assets are of national importance and systems need to be made fool proof, and potential rats kept away.

      • True; and in fact the word “cloud” is a bit of a misnomer … when you have PaaS, SaaS, IaaS, etc :)

        I think it’s useful though, as a concept, especially when speaking with business types. “On-premise versus cloud” is not a particularly useful or granular discussion in terms of IT staff, but in terms of businesspeople, it can help break them out of their on-premises obsession :)

    • A very valid comment in all senses.

      I think one of the things which is going to enable this kind of approach is freeing up mid-level IT staffers inside major organisations to trial things in their own little area. I think a lot of the use of Amazon Web Services, for example, came about because some mid-level IT guy got fed up with waiting for someone to provision a server for some trivial purpose and got a manager to OK a tiny Amazon contract for the same.

      We should be seeing these mid-level guys get lists of entry-level useful new applications from places like Gartner and testing them out for small, harmless solutions. This builds experience and shows what’s possible on a bigger scale. Then that knowledge feeds into bigger projects when they do get started. Right now, so many projects end up being SAP- or Oracle-only, I suspect because people don’t just have that much experience with other things.

  2. Yes lets put all our public data in the hands of “the cloud” and have our legal dept ensure that no individual decision maker is liable for security of said data. Then let us shed our knowledge base and skilled workers so that we can be done with having bridges back from where we came!

    I am a visionary leader among bureaucrats and so long as I can show a paper improvement until I’ve earned my hefty retirement purse all is good. I suspect my eventual predecessor will blame me for future ills but it will cost me naught. My son will at the least have a consultancy spot at IBM, Oracle, or maybe Booz Allen Hamilton… Jolly ho!

    • I don’t think anyone is saying there shouldn’t be legal responsibility in the right areas here; just that the old on-premises model was too heavy and too heavily customised to work as well as it was promised. The cloud model offers a way forward which allows many things to be trialled without the same commitment in terms of time and money. If things work, you move forward with them, but you avoid committing too much and getting bogged down in a massive IT project that is likely to fail.

  3. Cloud only? Right, so the same decision makers that screwed up all the IT projects are probably the same people giving the advice to these organisations. It may work in some sectors and technologies but not all. A hybrid situation makes more sense at this stage. But yeah good luck in trying to push cloud only.

Comments are closed.