The arrival of ASPs promises to uproot traditional software pricing methods. Nicholas Enticknap reports
The arrival of application service providers (ASPs) promises to solve one of the most pressing issues affecting IT sites. Software pricing has been a bone of contention ever since IBM started to charge separately for software in 1969, and the problem is becoming more acute as software occupies an increasingly larger proportion of the IT budget.
By submitting your personal information, you agree that TechTarget and its partners may contact you regarding relevant content, products and special offers.
In the ASP environment, the difficulties of pricing software according to its value to the customer disappear. Software becomes just a form of metered service, which all organisations have established procedures for dealing with.
According to Mat Hanrahan, a Bloor Research analyst, "Subscription-based pricing has a lot of advantages, aside from the difficulty of tracking client/server pricing. With the ASP model, you pay a subscription per month, per user, and delivery is up to the service provider. With client/ server, there is a big investment up-front and a big investment in implementation. You don't get that with the subscription model.
"An enterprise has to pay through the nose to ensure the system is secure. There's always going to be a hacker, but guaranteed security is what you get if you pay a subscription," he added.
Widespread use of ASPs is some time ahead, possibly five years, according to Hanrahan. But already software companies such as Microsoft are gearing up to adjust their traditional software pricing to the new world.
Pricing for server software is traditionally related either to the capacity of the machine on which it runs, or to the number of registered software users. Capacity-based pricing has a number of disadvantages:
- It means the cost of running a lightly-used application increases when you upgrade the host machine
- It makes it more expensive to run a small, new application on a larger machine than it does on a small one
- Pricing granularity becomes an issue: a small increase in workload can mean a big increase in software costs.
For these reasons, Isham Research analyst Phil Payne describes IBM's decision to introduce graduated capacity-based software pricing in 1990 as "the great mistake, that has driven so many small applications off the mainframe, and weakened the mainframe drastically".
IBM has taken a number of steps to alleviate this problem. It changed its Graduated Monthly Licence Charge (GMLC) system, where processors are divided into capacity bands, to a new system called Parallel Sysplex Licence Charge (PSLC), where pricing rises in small increments of equal size.
For users running new applications such as enterprise resource planning systems or Lotus Domino, it has introduced a new lower pricing structure, New Application Growth Environment (NAGE), to encourage users to develop these applications on mainframes rather than on Unix or NT servers.
IBM has also introduced usage-based pricing for applications that use less than 25% of the machine resources on which they run. The initial version of this had very little uptake, so IBM revised it and reintroduced it under the name Usage Licence Charge (ULC) in late 1998.
The company has promised that usage-based pricing will become more common in future and is trying to persuade third-party developers to introduce similar systems. Even so, it appears to be just as unpopular as its predecessor.
Mainframe software is essentially still charged on a machine capacity basis, despite the disadvantages. Although IBM tried to move the AS/400 over to user-based pricing in the mid-1990s, it had to move back again.
In the Unix and NT worlds, it is more usual for software to be charged either at a flat rate (typically for software on smaller machines) or on the basis of the number of clients licensed to access the server. There has been a trend towards introducing capacity-based pricing for larger Unix machines, in line with mainframe practice. Oracle recently offered the choice of either capacity or seat-based pricing, depending on which is more cost-effective.
Seat-based pricing raises the same questions of flexibility as capacity-based pricing: what happens when you go over a boundary and how you define a seat?
Ray Titcombe, chairman of user group IBM CUA, explained, "Browser-based applications may make full use of your back-end applications, but you don't have a client licence for the browsers. With e-business a client base grows rapidly to tens or even hundreds of thousands of users."
IBM decided that the best way to adapt software pricing to e-business was to develop transaction-based pricing. But, said Payne, "That's a long way off. It has proven too complex. One of the reasons is that they charge different amounts for different bits of middleware."
Another threat to traditional software pricing methods comes from the open source movement. This is the main reason for the popularity of Linux, which offers no technical advantages over other versions of Unix, but has the advantage that it is free to use.
Charging nothing is probably the only method of software pricing that arouses no resentment in any circumstances. Capacity-based, user-based and usage-based pricing all produce anomalies and difficulties. Moving to a service-based model has promise, but it is untried. It remains to be seen whether it solves the IT industry's 30-year-old problem.