Evaluation and Findings
The evaluation and weighting of criteria are based on interviews with twenty information technology (IT) academics and professionals, over fifty users, the management team of a bank, two SMEs, and three universities.
The universities and the bank both use combined solutions, while the SMEs use a Windows platform. The majority of IT interviewees agreed that the Linux OS is mainly server-oriented and Windows is mainly user-oriented, but these specialists are able to operate within both environments.
Considering hierarchy variables and market offers, it is clear that enterprises have three potential OS options to choose from: free solutions, commercial open source, and closed source. A combination of these three is also possible.
SMEs run critical business applications on their servers and workstations, and therefore it is important that they have some kind of guarantee. Free solutions offer no guarantees.
Companies usually have their own specific needs and must be able to decide what system they will use to meet those needs. However, when two alternatives offer similar capabilities, arriving at a decision may become difficult.
Last Part of the series Evaluating Strategic Information Technology Investment: An Appraisal of Software Alternatives for Small to Medium Enterprises.
Server Evaluation :
Universities tend to prefer free solutions, but this is not the case in the business environment. Universities focus on cost of ownership (CO), unlike SMEs. Other interviewees consider management costs (MC) and resource costs (RC) more important in the long term. Opinions are divided between these two criteria. While management places higher value on RC, IT professionals prefer MC. In general, these two variables are interdependent. However, all interviewed parties agreed that server platforms benefit from lower MC, while RC is more important for desktop environments. When evaluating categories of MC, IT professionals and academics agreed that system availability is the most important factor for server needs, while user-friendliness is the least significant. Patch management is the second most important factor, with weight on the installation rather than distribution. According to the interviewees, this variable influences security management. Opinions differ somewhat when considering software and hardware compatibility. Universities are more interested in hardware, while professionals give both factors equal value. For all those interviewed, deployment is the second least important value because it is not frequently required. Integration is by far the most important factor of the deployment category because of the constant need to incorporate an existing system with new hardware and software elements. Installation is the least important to all IT staff, because it is usually performed only once.
When evaluating RC, all interviewees stated that resource availability is of extreme importance because it influences all other variables. This variable includes human and system resources. To SMEs, the second most important factor is salaries, whereas for universities and IT staff, patch and drivers availability are the second most important factors. All interviewees agreed that centralized availability should be given more weight than forums and groups. It is interesting to note that none of the interviewed parties considered consulting and technical support as essential factors. Their explanation was that these two elements are unnecessary if adequate training has been provided. Unlike universities, which tend to use free software without warranties, the business sector assigns equal priority to the total cost of ownership (TCO) and warranty variables. One explanation for this may be that CO is much more important to universities than to SMEs, and that universities assign the highest weight to free solutions.
These findings suggest that because of the exceptional uptime, and therefore system availability, Linux is a better solution as a server platform (see figure 2). However, Windows cannot be discounted when considering server needs, as this OS generally performs better in the areas of resource costs, deployment, and compatibility. On the other hand, Linux has a lower CO, and it is the better choice in the areas of patch management, security management, and system availability (see figure 3).
Figure 2. Synthesis of the results for server evaluation
Figure 3. Sensitivity analysis of the results for server evaluation
Desktop Evaluation
For workstation applications, weighting values ddiffer significantly from those assigned in server evaluation. Both professionals and universities tend to use Windows at their workstations because almost every employee has a Microsoft OS installed on his home computer, which lowers the cost of training. Only the bank finds the warranty as important as TCO. Others deem TCO as more important, and assign it a higher weighting value. However, only two universities would consider Linux for workstations, as it provides a multi-OS environment.
CO is given the lowest weight in the overall model, while the IT staff prefers MC over RC. On the other hand, management tends to equally weight these two factors.
All interviewees agreed that the user-friendly factor is of the extreme importance. The consensus was that "the easier the system is to use, the lesser the costs and the greater the productivity." Costs and productivity are linked to system availability, which is the second most important variable. Since most of the interviewees do not use brand name workstations, they assigned a high weighting value to compatibility. In most cases, workstations are meant to run specialized software. Therefore software compatibility is strongly preferred over hardware compatibility. Patch and security management are almost equal in importance. An explanation for this is that effective patch management improves security management and vice versa.
Deployment is again considered the least important factor, since there is no need for multiple installation and migration. Therefore, integration into the existing hardware and software environment is assigned a higher value for the IT professionals. Aside from universities, the training variable is strongly preferred. When evaluating RC for the workstation hierarchy, the salaries category is excluded because the employee salaries are not linked with the type of system being used.
All parties interviewed find the patch and drivers availability variable of very strong importance. However, this variable is not essential for employees, but rather for IT and management staff. Same as with server evaluation, centralized availability is strongly preferred over forums and groups. Because workstations are maintained by well trained IT staff, and because maintenance of workstations is less demanding than it is for servers, the need for consulting and technical support is almost eliminated. Universities do not use it at all, and only one SME in addition to the bank uses technical support in an effort to keep IT staff at a minimum. All interviewees find the resource availability the third most important factor. This category has a different meaning for workstations and for servers. In this case, it primarily addresses the resources relevant to the users rather than to management, such as adequate literature, guides and manuals, and help personnel.
Analysis shows contrary results (see figure 4). Furthermore, when it comes to desktop applications, Windows outranks Linux (see figure 5). This may be because Windows dominates in the areas of technical support, training, patch and driver availability, flexibility, and (perhaps the most important) user-friendliness and easy-to-learn environments.
Figure 4. Synthesis of the results for workstation evaluation
Figure 5. Sensitivity analysis of the results for workstation evaluation
Combining the Best of Both Worlds :A few years ago, the question of which OS is more suitable solution would be irrelevant; Linux was less compatible, more complicated, and it had greater TOC than Windows. Today, it is clear that Linux has reached a certain level of maturity, and that SMEs can now consider its deployment on mission critical servers. In countries such as Germany, Nigeria, South Africa, and Korea, the government strongly encourages the implementation of Linux into business environments (see A. Grant's 2004 article Open vs. Closed source Software at http://www.scienceinafrica.co.za/2004/january/software.htm for more information).
It is interesting to note that free solutions do successfully compete with proprietary open source systems. The main advantages of commercial software are its warranty and better centralization of necessary patches and fixes. Also, there are no official professional courses and training related to non-commercial software.
The Windows platform is still the leader in the desktop segment, but it competes very well in the server arena as well. Moreover, Microsoft has advanced licensing policies with an expanding strategy that includes free or privileged licensing contracts for educational institutions, students, governments, and non-profit organizations.
This research suggests that the best alternative for most SMEs in terms of cost and benefits is to implement both Windows and Linux solutions; that is, Linux as a server solution with Windows used at workstations.
However, despite the results of this research, Microsoft is able to offer enticing conditions to its license policy, thereby returning leverage to the side of Windows.
Considering the above, the conclusion reached is that SMEs need to carefully collect and analyze all the available data in order to accurately evaluate their strategic IT investments.
