Mike Banahan, February 2006
This is a case study into the large-scale deployment of Open-Source desktops in schools. Two schools are studied, both using similarly designed systems. The software and hardware deployed is discussed and the management and cost implications arising are analysed. Both illustrate ways in which costs can be reduced whilst retaining compatibility with existing hardware and software investment.
The two schools studied both faced similar situations with tight budgets and much of their desktop hardware coming to or already at end of life. Schools of this kind are obliged to keep large number of desktop systems available to students: the Department for Education and Science (DfES) recommends a ratio of 1 desktop per four pupils.
The desktop operating systems in use previous to the Open Source deployment were mostly old variants of Microsoft Windows. To upgrade to modern versions of the operating system would require significant investment in new hardware because of the increased demand the operating system would make. Investment of that scale was difficult in the circumstances.
The schools adopted a comprehensive package of software, mostly Open Source based but with proprietary components. The solution has a large thin-client component which allowed existing hardware to be recycled and its life extended by a number of years. The administrative costs of the solution are estimated (with increasing validation available) to be very much lower than the equivalent proprietary fat-client alternative.
Software licence savings from the use of Open Source software give a significant reduction in system costs.
Reliability and availability figures indicate a much better level of service from the systems that was previously experienced.
Compatibility with software packages that will only run on Microsoft Windows has been retained where essential. In other places, Open Source alternatives are used. The Open Source software represents the bulk of the day-to-day use of the system.
The systems in use provide numerous features that were not previously implemented in the schools using proprietary software, i.e. an enhanced level of overall functionality compared to the previous situation. Some loss of functionality in certain areas is considered to be more than compensated for by those enhancement.
Very substantial costs saving (capital and running costs) have been indentified in both schools. Prima facie evidence of annual cost savings in the region of £100,000 to £120,000 can be shown in the absence of a more rigorous analysis.
The systems were introduced with low to moderate overall levels of disruption. In each case the system addresses only the teaching part of the school infrastructure and not the administration component.
Two schools were involved, each with a slightly different structure.
The organisational structure is considered to be of modest relevance overall.
In both schools there were deep concerns about ongoing costs and sustainability. Hardware upgrades on the desktop were driven by regular operating system upgrades on an approximately three to four year basis. Each time a signifcant operating system upgrade was needed, most of the existing desktop PCs had be replaced. For a crude estimate of costs, consider 300 desktops replaced every three years at approximately £500 each: that leads to some £50,000 per year cost in hardware replacement alone.
Replacement costs were not the only factor. Software licensing costs were of concern, particularly since record keeping tended to be inaccurate and it could not be guaranteed that every piece of software in use was accompanied by a valid licence. The costs of administration were seen to be onerous.
Orwell High School found that it was commonplace for several PCs in a classroom to be out of order for one reason or another and Sir Frederic Osborn had suffered serious problems exacerbated by the departure of technician support resulting in very low availability of working equipment. Not only was the cost of technician support proving to be high, availability of appropriate skills at the salaries that schools can afford was seen to be a serious problem. Addressing high-cost elements like these were critical decision-making points.
Though different in detail, both schools had broadly similiar initial circumstances. Both were previously using traditional PC desktops based on Microsoft Windows of various vintages. The cost of replacing PC hardware to keep up with the increasing demands of recent software releases was becoming unsustainable, but in each case there was a large inventory of low to middle specification PCs that could be recycled.
The effort of keeping PCs working was proving difficult to manage with considerable numbers of PCs awaiting attention to make them serviceable again. Meeting the one-to-four DfES guidelines on PC-to-pupil ratios was considered to be unrealistic at Orwell. Sir Frederic Osborn had substantially exceeded that ratio prior to its attempt to upgrade its systems during Summer 2005 however
"This came at a great cost - the IT support levels were not increased to underpin this ratio. The consequence was a steady reduction in reliability and increasing systemic problems; most of which were derived from the Windows fat client approach. Also the demands on the network were substantial."
Prior to adopting the system, both schools trialled a demonstration system for a limited period.
Orwell High School was the guinea pig for much of the development work on the system, adopting it before it was packaged as a commercial product, and consequently could be seen as taking a greater risk. In practice there is known to be several schools in the UK using systems that have similar characteristics to the one in question, so the technical risks were viewed as modest.
Sir Frederic Osborn School was able to see the system in use at Orwell after a year and almost a further full term had elapsed, permitting a high level of confidence to be established. The school had been looking for a solution similar to the one adopted but had ruled out a self-built implementation. They were able to choose it when they discovered that commercial support had become available subsequent to work done at Orwell, having seen a story about it on the BBC 'Working Lunch' programme.
The system implemented contains numerous components. The most significant element is the thin-client Linux-based part, but these are augmented by others. In brief the system provides:
Thin-client systems have some known limitations when applied to heavy multimedia content and these are no exception. Streaming media such as delivered as part of online web pages is well supported and specialist multimedia is better addressed by the use of specialist suites of equipment which falls outside what is offered by the system under study.
The ability to use recycled PCs has helped keep costs low. Orwell equipped an entire classroom with PCs junked by a bank for the cost of a minibus sent to collect them, a figure of £40 in total being mentioned.
A full description of the features and components of the system at this point would be lengthy and disproportionate. Interested readers are directed to the references section of this document.
The schools differed in their approach, so each will be treated separately
The system was installed during the summer break of 2004. Few staff knew about the installation and for most it was a surprise when they returned. Pupils knew nothing about it until they first encountered it. It is acknowledged that the surprise 'big bang' approach is less than satisfactory, but apart from some grumbling from members of staff who had to adapt lesson plans to suit different software and a small amount of initial discomfort from older pupils who knew Windows well, the system is reported to have settled in with little disruption.
Due to difficulties with the previous Windows systems, staff and pupils were keen to obtain access to the new system. News of its installation was well-known in the school, by contrast to Orwell. Limited staff training was undertaken (MORE TO FOLLOW).
Determining an annual figure for costs is difficult and few schools conduct exhaustive analysis of the combination of likely running costs in terms of the important variables, amongst which are:
Accurate cost analysis would require detailed study and would take more time than is available for this case study. It seems clear that there is substantial evidence of potential savings and a rigorous analysis would be highly desirable. A figure of £98,750 per annum per school is persuasively arguable in the light of the figures shown above.
An internal cost appraisal undertaken by Sir Frederic Osborn school showed a predicted cost saving in the first year of some £120,000 compared to the alternative, which would have been large-scale replacement of almost the full desktop hardware and software suites. Tim Lee, who evaluated the costs adds:
I should note that my cost analysis shows the biggest cost savings when the full support requirement for all 420 machines is factored in. I did this at the rate of £25k for junior and £30k for senior techs (this includes all ancillary costs it is not only salary). I assumed we would need 2 additional junior techs to support the full complement of Windows PCs. The fact is we could never afford both, though we probably could have afforded one extra. The figure of requiring 4 techies [assuming 420 machines] came from multiple sources so I am confident it is correct. We are hoping to reduce to 1.4 (FTE) techs, though my cost analysis allows for 2. I think we are likely to have to stick at 2 to provide adequate cover etc. The additional .6 is likely to be a student on work placement - we have had this for the last two years and it is usually very cost effective. The transitional costs, particularly on network setup, have been higher and more complex than originally expected.
Users note a dramatic increase in reliablity and availability of the desktop systems. The software packages are more than is needed for mainstream use and provide good compatibility with their proprietary alternatives. Training/classroom materials have needed to be adapted to the new software but this often also happens when upgrading from different releases of the proprietary software packages. More training during the switchover period would be beneficial but in general the system performs well, meets the needs of the organisation and is not a problem to use. Overall it is considered to be an improvement on what went before it.
In both schools there have been organisational issues which precluded interviews with systems administration staff.
The installation at Sir Frederic Osborn highlighted the need for proper project planning during commissioning of the system; time was wasted in overlap between electrical provisioning, commissioning of the network and server and desktop system preparation. Use of Gantt charts and a project plan would have been beneficial but was not done.
The management of both schools see the system as a long-term solution to some intractable problems of demand that cannot be realistically be met within existing budget constraints. At Orwell High School the system has been running for well over a year and has met and exceeded expectations: there is no doubt there that the decision was the right one to take. At Sir Frederic Osborn the system is relatively new and so less can be said about it with confidence: there, it has at the very least, solved a very serious problem of delivering reliable computing to the desktop.
Both schools saw the availability of external commercial support as very important. Whilst other schools have employed a D.I.Y. approach to building similar systems, the risk of losing the technical skills to support the system if a key person leaves is high in those cases. An externally designed and supported system is seen as a more desirable solution.
Sir Frederic Osborn School adds:
"Students and Staff have both observed and been delighted with an immediate increase in reliability and speed. The increase in speed is kind of surprising considering there are now up to 30 people using one server"Sir Frederic Osborn School has also asked that mention of the importance of the network is made in this case study. A thin client deployment is highly network dependent and weakness or unreliability in the underlying network becomes visible immediately when it is used in conjunction with thin clients.
The two schools at the core of this study both now have reliable, functioning classroom delivery of ICT. That might be taken as a 'given' but the difficulty and cost of delivering a reliable service was previously a problem in both establishments.
Both schools say that they now have a sustainable model which can be managed within their existing budgetary framework and one which provides a platform upon which they can build.
Switching from a proprietary Microsoft Windows platform to one which is predominantly Open Source inevitably implies change. More attention paid to training administrative and teaching staff could make the switch somewhat easier but overall it is felt that the new system provides a better experience, the improvements outweighing any disadvantages.
The analysis of costs and savings is complex and difficult. Historical figures can be hard to determine and future costs for proprietary software are generally guesses only. Where savings have been achieved it may be argued that some are not inherently related to the use of Open Source Software in particular but are equally due to reorganising and better managing service provision. A detailed analysis would probably take some months - but in view of the potential savings for the nation, the author believes that government would be remiss in not undertaking such a study.
No appendices are included in this case study.
These references are not necessarily supporting the contents of the study, some are included for further reading.
All trademarks are acknowledged as the property of their respective owners.