Office moves

There are two fairly immediate actions that organisations can take if want to quickly reduce their reliance or their outlay on desktop PCs running a stack of resource hungry Microsoft applications.

They can replace those applications, and their desktop operating systems, with open source software equivalents, thus reducing licence fees; and, they can re-architect their whole systems set-up, focusing much more on server-based computing, which should reduce support costs.

It is possible that neither approach will yield the returns or the control promised. Nevertheless, many buyers feel they cannot wait to see if Microsoft’s .NET (see article, Battle for control) reduces overall management costs, nor can they ignore the prospect of large-scale savings and greatly improved control. They are exploring their options right now.

There is no unambiguous research that suggests a mad rush to either server-based office computing, or to open office solutions. But clearly, an increasing number of IT directors are beginning to feel something like exasperation over the constant stream of problems that beset their PCs, whether it is security, user mis-management, or simply upgrade costs. And that exasperation is beginning to show.

“A PC represents a severe security risk because it has local storage, a local Internet connection, and local media drives, all which can be channels for viruses ” explains Chris Wood, IT manager for UK-based logistics company MAT Group.

Another IT director complains: “With each software upgrade comes the possibility of a renewed hardware investment. For example, upgrading from Windows 97 to Windows XP requires a completely new set of PCs with faster processors, more memory and bigger storage space.”

The explosion in the use of personal digital assistants (PDAs), ‘smartphone’ mobile phones, and laptops, which can be less reliable, less compatible and are easier to damage and lose, has complicated matters and, for some, made a move to less expensive or more easily manageable alternatives more attractive. Around 42% of UK corporations own between 100 and 500 laptops, according to market research from UK-based Total Romtec.

All of this – along with Microsoft’s controversial new Software Assurance licence scheme – has brought the question of using one of the two alternative architectures to the fore.

Server-based approaches

The first and most widely adopted alternative to the so-called ‘fat client’ is to use server-based computing. This involves centralising all or most applications, data processing and storage on servers – an approach that has already proved itself effective and robust for business usage. Not only is this architecture flexible and easy to maintain, it also provides a foundation for delivering Internet-based applications. Suppliers such as PeopleSoft and SAP have made much of their move away from client/server computing to simple server-based systems, in which all applications are accessed through a browser.

Server-based computing is a confusing term in the context of office and productivity applications. Most server-based applications are big applications, and they are run on servers for good reasons: they are usually complex, require a large, scalable database, require considerable management, are accessed by many people or systems, and are ‘mission critical’. Financial, manufacturing or line of business applications are typical examples.


Alternative architectures to the fat PC

Client-server: Two-tier and three-tier client-server architectures consist of application logic stored on the client device, accessing data from a central database, perhaps with a third layer of centralised processing for more demanding applications.

Server-based: There are three types of server-based computing models.

Thin-client: this uses a very small piece of client software to receive and display applications based on a server, transmitted using proprietary compression techniques over a variety of different network protocols; Web-based: similar to thin-client but using a web browser as the client to view applications based on web servers, which are transmitted through the standard Internet protocols.

Dumb terminals:Which carry no client, have no processing power and no storage, but can display applications in graphical user interface (GUI) format transmitted directly from the server over proprietary protocols.

Rich-client: Uses client platform to store chunks of business logic and data, such as role-specific settings, to enhance and speed up the delivery of server-based applications. Client software can be downloaded form server in each instance of use, or stored on desktop as a recent cache from the server.


Office applications, such as Microsoft Office or its alternatives, are not obvious candidates for running on servers. Most of the processing, and the data, is used by an individual; the data is neither shared, nor is it usually mission critical. Indeed, server-based office applications have been developed over the years by suppliers such as Cincom, Wang and ICL, and all eventually failed because users preferred the control and immediacy of local applications.

But there are two reasons why this might be changing. First, around half of an average business’s IT budget is spent supporting and maintaining desktop client architectures, according to Ashim Pal, analyst at market research company Meta Group. This is an enormous figure – considering that European companies’ IT budgets represent 2.89% of their revenues, according to Gartner Research. This could rise to 4.07% by 2005, claims Gartner.

Second, infrastructural and technical changes have made centralised systems more practical and more desirable. Local and wide area networks (LANs and WANs), for example, have become faster and more reliable, while the wide variety of end-user devices makes centralised files more desirable.

To meet this demand, many different models for server-based office computing have been developed (see box), each offering a variation on a common theme – most of the processing, storage and management is done centrally. There are also different methods of delivering these technologies: companies can implement server-based desktop applications within their own data centre, or access applications as a hosted service from an application service provider (ASP), such as the UK-based CSF, or BT Ignite.

One thing is clear, however: Despite a growing number of alternatives to the Office suite of productivity software, Microsoft applications are still the preferred standard within the enterprise. As Gartner’s Silver puts it, “Microsoft Office is part of the basic IT infrastructure of the enterprise and cannot be eliminated or easily replaced.”

Indeed, few IT directors – at least until the controversial new Software Assurance licensing scheme was introduced in July of 2002 – have wanted to get rid of Windows or Office per se. Their main priority is simply to reduce the cost of supporting a client-centric desktop architecture.

This has led to a rise in demand for products from companies that specialise in running office and productivity type applications on the server side – most notably Citrix and Tarantella. These companies supply software that enables Microsoft Office applications to run on a server, and be accessed and worked on by users working with web browsers or on thin clients. (Microsoft also offers this with its Terminal Services product).

“One of the main reasons people choose us is to be able to amalgamate systems so that they can publish to any platform and device,” says Keith Turnbull, general manager of the mobility group at Citrix, one of the leading thin-client technology vendors. By doing this, vendors like Citrix claim that customers can rationalise their software products and versions, and will find it easier to centrally archive information and incorporate it into knowledge management systems.

The advantage: all server-based applications and data can be managed centrally, meaning much lower support costs. That means viruses, security, back-ups, upgrades and help desk and problem resolution can all be handled without ever touching a desktop machine.

The success of this approach can be seen in the financial results of Citrix, Tarantella and others. Citrix’s annual revenues are $591.6 million (EU600.5m), while Tarantella’s are $67 million (EU68m) – impressive figures for young companies. But those same sales figures also help to put it context: Microsoft earns about $10 billion a year from sales of Office (including server-based Office). In revenue terms, the proportion of Office users supported by a server amounts to a few per cent.

Why has it not been more popular? There are several reasons, including the fact that Citrix or Tarantella licences add extra costs: a Citrix licence will cost almost $6000 (EU6089) for 20 users.

But there are other reasons. “No one wants a dumb terminal because they have got used to all the extra applications, such as games and media players, and the ability to personalise their desktops, which is included in client operating systems,” explains Andy Bovingdon, UK-based director of product marketing for Tarantella.

As trivial as it sounds, this is a common reason cited by technology vendors to explain why server-based computing is yet to be adopted widely. But there is, perhaps, a more fundamental problem with server-based desktop computing: a constant reliance on a network connection.

Unlike client-based applications, the responsiveness of server-based applications is directly related to network bandwidth. Unless high-quality bandwidth is available, application responsiveness cannot be guaranteed. This not only affects mobile workers, such as sales people and site technicians, but those within the enterprise’s four walls as well.

Giga analyst Mike Gilpin cites one bank that claims it saves $1 million (EU1m) for every three seconds shaved off its call centre response time. For this reason it chose to implement client-centric software over a server-based application, which is more common in such an environment.

For reasons such as this, thin-client or server-based architectures work well at a ‘heads down’ business, where network connections are immediate, as opposed to one with remote or mobile workers, explains Ashim Pal, at Meta. ‘Power users’ who want regular access to the full suite of Microsoft applications may not be, and may never be, suitable customers for server-based computing.

A complication of this issue is that there is little point in eliminating the most desktop-friendly application – Microsoft Office – if other applications require a rich-client PC and all the support that this entails. And herein lies a further problem: “In many cases… thin-client applications have not matched the ability of ‘rich-client’ to deliver superior performance and desktop functionality,” says a recent report by the Aberdeen Group. “Especially in small to medium-sized businesses, or workgroup or departmental applications, the results are failed conversion efforts and customer-offending slow performance.” Aberdeen counsels users to make sure they invest in the right tools to ensure thin-client applications work effectively.

Meta Group analyst Ashim Pal warns: “The server-based model turns the complete infrastructure upside down. You may need a huge server farm and a fat network”.

None of this likely to stop the spread of server-based computing. Although the research is not conclusive, analysts are upbeat about the appeal of server-based computing. A Giga Research Group survey finds that 62% of businesses have deployed server-based computing, while 25% consider the architecture as strategic to the enterprise (see box).

According to Giga Research Group analyst David Friedlander, the Windows server-based computing industry will grow by between 10% and 20% annually until the end of 2004, when it could become a $2.9 billion market. Such research, however, appears to take a much wider view of server-based computing.

At best, says Meta Group’s Ashim Pal, server-based desktop architectures will eventually account for 60% of an enterprise IT architecture, and this will not be for a “very long time”.

Meanwhile Infoconomy/Information Age‘s own reader research, based on responses from over 350 user organisations, perhaps puts the move to server computing in a more conservative light: Only 7% say that they are delivering office applications over a network to thin clients as an alternative to PCs; 40% say they have not even considered it, and 18% don’t know.

Open source

In February 2002, the Association of Danish Municipalities, representing Danish local government IT directors, began investigating the use of alternative desktop software packages for its 55,000 desktop computers. The reasons: to avoid the price hike expected with the introduction of Microsoft’s Software Assurance licensing scheme.

In July 2002, the UK government followed suit; it said it would consider open source software as an alternative to proprietary software such as Microsoft’s Office. Other European governments – including the German and Swedish governments – are said to be working on similar proposals to use open source software.

In the private sector the news has been just as bad for Microsoft. One of the few private companies to go public about switching away from Microsoft is Verizon, the US telecommunications company. It recently revealed that it had saved $6 million by replacing its programmers’ Unix and Windows workstations with Linux systems that run OpenOffice, an open source office productivity software.

These organisations are not alone – Sun, for example, claims there is strong interest among customers wanting to drastically reduce their Windows licenses – especially where they have a lot of desktops and where they do not upgrade regularly. On the whole, licence fees for desktop alternatives, such as Sun’s Star Office or OpenOffice (see article, Office and its rivals) are less than 15% of Microsoft Office. In some cases, they are completely free.

Open source software, such as the Linux operating system, has gained traction over the past decade as a server-side application platform, because it is freely available, easy to modify, and robust. It has, however, had far less impact on the desktop. Microsoft’s new licence programme, however, appears to have triggered renewed interest in open source software as an alternative to desktop operating system and software packages.

For example, Microsoft’s chief rival, Sun Microsystems, recently released 5 million lines of StarOffice code as open source software, and plans to sell Linux for some corporate desktop computers. Meanwhile, alternative desktop applications suites, such as the Gobe Productive office suite – originally used for the now defunct BeOS operating system – have also been turned over to the open source community.

Companies such as US-based Ximian specialise in turning free open source desktop applications and operating systems into enterprise class products – it is even working on an open source alternative to Microsoft’s Common Language Runtime (CLR), a key component of .NET. Others, such as Lindows and CodeWeavers, are working to make Windows software run on Linux machines.

The success of open source is demonstrated by a survey of chief information officers in March 2002, by the open source organisation OpenForum Europe. Thirty two per cent are looking at open source software as an alternative to reduce license costs, 39% currently use open source software, and 49% plan to do so in the future. However, while open source is broadly deployed on servers, only 14% of CIOs consider that it will have a role on the desktop.

Analysts caution that the open source products have a way to go before they can compare with the quality of user experience of Windows. And, moreover, there is a cultural aversion to using free, or almost free, software. Many IT directors worry about support, future development, quality and liability for problems and, as John Handby, CEO of the CIO Connect, a membership organisation for IT directors, points out, “Linux is only free if you regard our time as worthless, due to the support and set up costs.”

Currently none of the alternatives to Microsoft Office can act as a complete replacement to Office. This is because they lack the full functionality of Microsoft’s suite. Furthermore, it is not always possible to open Microsoft Office documents in StarOffice, although this is said to be less of a problem in the latest version of StarOffice. This is a major worry for customers.

Yet, alternatives such as StarOffice, which, for example does not include a database programme such as Microsoft’s Access, can still be suitable for certain users, says Gartner’s Silver. He says that most corporate desktop users do not use 20% of the application functionality provided by Microsoft Office.

His advice, therefore, is that organisations that have a deep understanding of their users’ requirements can migrate low-end users to an alternative to Microsoft Office – although incompatibility issues will need to be be overcome. But, in contrast to Sun’s claims, Gartner argues that the cost of migrating, maintaining, and training users for StarOffice are likely to outweigh the drop in up-front license fees.

Skills and training is a key issue: programmers or technical support staff expert in Windows may not easily adapt to a Linux-based environment. Of course, adopting open source office applications is only a tactic: the problems of desktop support and user management can be just as great with open source applications. For that reason, some customers will combine the move to open source with a move to server-based computing.

In spite of all this, many, and probably most, customers won’t move away. All this assumes that desktop software itself will not change significantly. But there is a strong chance that it will.

Microsoft, whose products are universally client-focused, has seen the slowly gathering migration from the client to the server, and has reacted to protect its interests. A key element of this, is, of course, .NET and the related and controversial Software Assurance licensing programme.

“We have not yet decided the balance of what is on the server or client. .NET takes into account the processing power of the clients so developers can make web services applications thin-client or desktop based”, said Gavin King, Microsoft’s .NET spokesperson.

Ultimately, says Gartner’s Michael Silver analysts, at least half of all organisations will have an infrastructure that combines a number of heterogeneous computing models, including server-based, and rich-client architectures, and will run any number of different operating systems for different desktop requirements.

Pete Swabey

Pete Swabey

Pete was Editor of Information Age and head of technology research for Vitesse Media plc from 2005 to 2013, before moving on to be Senior Editor and then Editorial Director at The Economist Intelligence...

Related Topics