Today, businesses have more options and more flexibility in acquiring licensing to accommodate their software needs. However, with that flexibility often come complex software asset management (SAM) obligations, the neglect of which can prove to be extremely costly.

With regard specifically to software installed on servers, accurate licensing can involve a broad array of options, because server applications by and large perform a much broader array of tasks than those performed by workstation applications. In practice, though, most server software is licensable only under one or two of the following models:

  1. Server Installations and Clients. With server licensing, server + CAL licensing or processor licensing, the licensing "math" is relatively easy in comparison to other server-licensing models, though for many products, it can still be difficult to determine the client-count cutoff point where it becomes more cost-effective to purchase processor licenses as opposed to server licenses and CALs. In addition, the flexibility and scalability of this approach usually is limited.
     
  2. Processor Capacity. Under this approach, the software publisher assigns a processor value to a make and model of physical processor chip or processor core (per IBM, a "Processor Value Unit" or "PVU" per core value). The per-processor value then is multiplied first by the number of processors or cores running on the server and then by the per-value-unit price set by the publisher in order to arrive at the total license cost for the server. This model usually allows for a greater degree of flexibility and cost savings for businesses that deploy virtual servers or logical partitions on physical machines. However, such sub-capacity licensing typically involves very complex licensing arithmetic and, in some cases, additional licensing requirements related to monitoring the server resources actually accessed by the software.
     
  3. Function-Specific Capacity. The wide variety of functions addressed by server applications is mirrored by an increasingly wide variety of function-specific licensing metrics for those applications. For example, data-backup software might be licensed based on the volume of data subject to backup operations, or database software might be licensed based on the maximum number of concurrent connections to the database during a reporting period. These models have the advantage of allowing for licenses to be purchased on a more incremental basis as functional needs grow over time. However, function-specific license requirements can be difficult to measure.
     
  4. The Cloud. As they do with workstations, many businesses also look to hosting services providers to deliver server-based application solutions over the Internet. This can carry advantages in flexibility and scalability, but it also usually raises significant concerns related to third-party control of IT resources.

With regard solely to acquisition costs, the ideal server-licensing solution typically involves some combination of the various licensing models identified above. However, correct license-counting in such a "mixed" environment can become very difficult and fraught with challenges. There are few points along the process where corporate counsel cannot make positive contributions to the implementation of appropriate licensing options, and there are certain points were legal involvement is crucial to ensuring that the company remains compliant with applicable license agreements.