Cost cutting in IT (Information Technology) has many possibilities. Hardware and Software costs are the primary ones. Hardware costs came down over the years, as per Moore’s law. For a given unit cost of hardware, higher performance was provided over the years. The hardware costs have got reduced so much that computing costs and power costs have become comparable. There has been greater focus to also deliver more computing per unit of energy (The Price of Performance, by Luiz André Barroso of Google, Multiprocessors, Vol. 3, No. 7 - September 2005). Hardware in general is not made to the individual’s requirements. However, enterprises could choose their hardware from a number of options available and make it suitable to their need. Similarly the network / bandwidth costs decreased over the years.

Interestingly enough, software costs have not come down over time. Perhaps they have increased. Vendors have continuously innovated and kept up their prices over time. Web servers and browsers were sold for a premium during the initial days of World Wide Web. When they became commodity, applications servers came in, as part of the evolution. Similar parallels could be taken with Microsoft Office type products. Open source communities have constantly provided old technologies for free. The Open source software gained momentum and became a part of the mainstream. This has helped the cause of cost cutting or control with respect to software.

If software can be classified into functional and non-functional (or infrastructural), the latter variety has been the target of open source. Linux operating system, Tomcat web server, Eclipse platform, JBoss application server etc, are various examples. However, functional software that caters to the needs of business: accounting, supply chain management, customer relationship management, retail banking etc; have not been part of the open source movement. It continues to be expensive for various reasons. Two banks in a geography will provide differential services or offerings to stay in business. That is incorporated in the software used by them. Offerings of a given bank would change from geography to geography. Software developed for a bank can't be used by an insurance company. On the other hand, non-functional (or infrastructural) software can be the same for all of them in all geographies. Therefore, formation of open source community for/and development of functional software is not same as the case of non-functional (or infrastructural) software.

Functional software could be classified further into Packaged Applications and Custom built Applications. Couple of decades ago entire functional software was perhaps custom built by enterprises. As time progressed, software companies started making functional software upfront and configuring the same for an enterprise. Examples are SAP, Siebel, Oracle Applications, type of software. These Packaged Applications helped in reducing implementation cycle time. Broader acceptance has also helped in standardizing skill set. While the fact that Packaged Applications have helped enterprises is undisputable, whether they helped bringing down the costs is debatable. The Packaged Applications vendors have tried to the address the ‘cost’ part progressively. However, these applications have not received the attention of Open Source community. Hence the cost of these packaged applications did not decrease like in the case of non-functional software.

Custom built applications continue to be expensive. Life cycle duration and manpower costs are perceived to be a lot high. To address the life cycle part – a number of methodologies or models have come into place. Be it Waterfall or Iterative or Agile methodologies. To reduce manpower costs, outsourcing / offshoring have been adopted. Offshoring has triggered development approaches to adjust to distributed teams. The development tools, infrastructure type of tools (Citrix, VMWare, …) have aided distributed teams. Once these methods of distributed development and infrastructure become mature and easily affordable in future, teams could be anywhere on the globe, where quality manpower is available at competitive price. Open source communities have thrived on distributed development techniques. To replicate the same for software development within an organization or extended organization (including offshoring / outsourcing partners) has not been easier. Data privacy, intellectual property and security requirements are stringent in these cases. However, distributed development techniques as well tools will continue to mature and help in containing manpower costs.

Ability to use ready made products / items as is will help in standardization and commoditization and thereby reduce costs. One example, readymade clothes vs custom tailored clothes. Wherever custom made items become a premium, non affordability will drive consumers to accept ready made items. Same could be said about owning an independent house versus a flat. Perhaps SaaS (Software as a Service) is for those who can’t afford custom software but still need it at reasonable price (ABC: An Introduction to Software as a Service, by Meridith Levinson, at www.cio.com, May 15, 2007). Other way to see is – all functional software pieces with diminishing differences and providing no competitive advantage to enterprises will be available for cheaper costs in future.

October 2007