Open source software might be a hot topic among government agencies, but it’s still not for everyone.
Open source software (OSS) is now one of the hottest technology interest areas for government, both in Australia and around the world. Media attention has perhaps raised this to the point of it being “hyped up”. But how much potential does OSS offer for achieving real, tangible benefits — or is this mostly speculation that will stir a lot of attention but not really deliver?
There are two prime perspectives that governments can take about OSS. The first is a pan-government policy perspective. OSS holds the promise of delivering aggregate cost savings compared to acquiring and using proprietary software, probably the strongest driver of interest for Australian governments. There is also a strong desire by some governments to reduce their dependence on large proprietary software providers, most notably Microsoft and their desktop products. There is also an attraction over the potential to grow a local software development and support industry through the use of OSS. These considerations are especially strong for governments in “developing” Asia/Pacific countries such as China and India, but also hold true for Australia.
The second key perspective for OSS is with government agencies as individual buyers and users of software. This column focuses on this perspective, particularly in identifying where OSS fits best and is likely to yield substantial benefits for agencies.
It is perhaps revealing to see how widespread the use of OSS is already. Gartner surveys show that some 48 per cent of government organisations in Australia now use OSS, either directly or through the IT services they source. By comparison this is much higher than for enterprises in the financial services and manufacturing sectors, but lower than for the “high OSS” sectors of Communications and Education where usage rates are 80 per cent or greater. Overall in Australia, some 52 per cent of organisations currently use OSS.
This may surprise you. That’s because the major usage areas may not be all that visible to many agencies. OSS has proven to be strongest in IT infrastructure areas, providing critical building blocks for many “edge of network” uses, such as proxy serving, caching, VPN and the like. This means that its usage is quite ubiquitous but attention is not top-of-mind, especially as its success has been founded on a high level of reliability. The best-known example of this is Apache, the leading product in the Web server market.
The next strongest area of usage currently is with “compute clusters” in areas such as financial functions, research and development, biotechnology, geophysical, energy, and applications such as visualisation. The Bureau of Meteorology (BOM) is a notable Australian example, having developed core enterprise applications for weather monitoring and forecasting, based on Java and Linux. It did this through collaboration with other weather organisations and a group led by the University of Wisconsin in the US. Importantly, the impetus for moving to OSS came from senior management in the agency, not from technologists.
So it is useful to view OSS as being a disruptive technology. Not that it’s all that new — the core of Linux originated decades ago. Early adopters in government have typically used it as to substitute for proprietary software and development alternatives in niche areas where special performance attributes are needed. Next adoption is then likely where OSS can compete successfully in other roles, especially where high reliability is required. Later adoption would focus on providing low-cost solutions that can deliver lower price and risk, where knowledge and skills needed for implementation and support are widely available.
For mainstream enterprise usage, OSS adoption now is strongly under way with major IT vendors and service providers. This movement is powerful, with commitments declared by IBM, Sun, SAP, Oracle, EDS and HP. The notable exception is Microsoft, the highest profile proponent of the commercial software model and arguably with the most to lose from a widespread adoption of Linux.
OSS on the desktop — which in practical terms means Linux — raises a different set of considerations. Agencies considering migrating their desktop PCs to Linux must weigh factors that go beyond hype and any anti-Microsoft sentiment. These factors include the composition of your application portfolio, the requirements of your users and the all-important migration cost and return on migration investment.
Spending on a major migration that won’t show an ROI within two to three years usually does not make sense. But there are situations where a move to Linux OS on the desktop will deliver ROI and does make sense, but it’s in a very narrow, limited range of situations. It should be considered if there are relatively few applications, with these being fixed-function or low-function, such as data entry, call centre or kiosk uses. The cost of migration may be low enough to justify the move to Linux.
PC vendors have so far not been enthusiastic about Linux on the desktop, using it only sporadically in sales campaigns to meet specific price targets. Overall, Linux support on the PC is the exception and not the rule, an important limitation for an agency. If you don’t have these skills or if support is outsourced, flag the need with your service provider. But if they also don’t have the requisite skills, then risks are very high.
In practice, most agencies will probably face the question of whether to migrate to Linux OS when they replace desktop PCs where they usually face upgrading both Windows and Office software. This issue requires detailed analysis of the costs and benefits of moving from the early version of Windows now in place, not only to Linux but also to the latest Windows version. The most viable alternatives to Microsoft Office, StarOffice and OpenOffice, offer less functionality — though this may be adequate for many agencies — but, importantly, there are compatibility issues with some Office-created files and macros.
Organisations have three core concerns with OSS becoming mainstream. The biggest one relates to the risks involved in its integration with pre-existing legacy systems and packaged software applications. Another is with the availability of skills needed for development and support. A third relates to the training of staff who use any new OSS applications. These are valid concerns and need to be specifically addressed by each individual agency.
So OSS is rather like the curate’s egg — there are good and bad bits. Here are some final pointers for where OSS is likely to fit best.
- Where an OSS product is especially “suitable for the purpose” and offers good value for money. Use OSS where it is strongest, especially at “edge of network” situations and in non-competitive cases where collaboration is important (like the BOM example).
- If OSS products are close architecturally to your current environment and skills that you already have (for example, for internal applications development: skills with Unix, Java and Web services).
- When OSS products are a less risky and more viable option to internally developed applications. But be careful — many products are not yet mature enough for critical usage.
- Where visibility and repeatability are important, such as large-scale deployments (for example, education projects; technology dissemination among wider populace) and especially at the server end.
- Linux will generally have better price/performance characteristics against “simple” RISC/Unix configurations, and against Windows in larger, replicated server placements.
- And perhaps one of the most valuable attributes of OSS arises from using it as a competitive negotiating weapon.
And a parting thought. Government CIOs should especially understand the politically-charged considerations and debates around OSS. Because of vested interests, they are likely to be scrutinised more closely in the future. Be well prepared.
Richard Harris is vice president of Gartner Asia-Pacific