Computing - Server
Multicores, Virtualization, and the Server Software Licensing Conflict
February 04, 2005 / Sagitta Pan
7 Page, Radar

Abstract

Throughout the history of the IT market new technologies have typically played a major role in driving growth, while at the same time challenging the existing business models of industry players. Similarly, the current march toward multicore processors and virtualization technology proposes to bring substantial benefits in stimulating market demand; however, this building trend threatens to generate soaring licensing costs. The impact on operating systems and software for high-end servers will be particularly strong. This Radar will examine these new server technologies and identify the problems and conflicts likely to arise from software licensing models, as well as possible resolutions to these conflicts.

Given the intense state of competition among server chipmakers, the implementation of multicore and virtualization technologies are unlikely to be phased by software licensing issues. However, ultimately the key to resolving the licensure dispute sits in the hands the software players, and at present there appear to be three separate models taking shape.

The Computing Power Model

Processor count was previously the chief way for software developers to measure server computing power. With multicore technology introducing higher processing power in the same number of chips, forcing software players to base licensing structures on the number of physical processors is not appropriate. Truly, the crux of the matter lies not on the number of processors, but rather, total computing power. 

If system makers are able to offer a way of measuring computing power, this will be a major step in helping software developers conceive a computer power based licensing structure. Such a mechanism would be most applicable to enterprise mainframes or servers for scientific applications. For example, UDC (Utility Data Center), promoted by players such as HP and IBM, leases computing power to corporate clients -- a model that would be applicable to software licensing schemes as well.

If software and hardware makers are able to base licensing mechanisms on computing power, enterprises would be afforded a "use as you need, pay as you need" payment model. This would fit well into the utility computing concept, and the utility computing business model could develop unhindered by software licensure issues.

The Contract Model

To address licensing issues in the small and medium server or personal computer segment, software providers can adopt an IT services model. For example, Sun's Java Enterprise System calculates annual service fees based on the number of enterprise personnel, rather than outright sale of software as previously done. The streamlining of what could be a highly complex licensure model and the lowering up-front procurement costs would increase the willingness of enterprises to implement the new technologies.

This model is similar to the flat-rate charge schemes of Internet access service. Varying levels of service are offered for time periods required by the customer, while the amount of usage is not a factor in pricing.

The Outsourcing Model

Enterprises could also go the route of negotiating for the external provision of IT hardware and software, and avoid the computing power and software service models altogether. At present the IT services market is served primarily by IBM, EDS, and Fujitsu, and with these companies handling software and hardware, enterprises can avoid dealing directly with licensure issues.

Establishing a business model to address the contentious licensing debate will be critical. The sales method used with the traditional product ownership model of before will not sit well with the service or task-oriented IT service models that are becoming more prevalent. To solve these issues, IT players can either use computing power for task-based calculations, or fees based on service grades. This will not only solve the present software licensing conflicts, but will also support the advancement of utility and grid computing.

Recommended Research Reports

>No Data
To get MIC's complete insight, please log in.