These comments are supposed to draw a comparison between today’s software as a service (SaaS) market and another, extinct software delivery model with some important similarities and differences. The purpose of that comparison is to provide insight into the modern SaaS market, while summarizing the history of an important era of computing.
A Little History
A computer’s cycles were once divided between many users, a practice known as timesharing. A mainframe would have been leased to customers of a data processing service (DPS) in terms of processor cycles, storage per unit time, and terminal connections. The first timesharing systems were developed at MIT and Dartmouth in the late ‘50s and early ‘60s, and commercialized by General Electric GE in 1965.
DPS companies also provided batch processing services, the non-interactive processing of a customer dataset. In general, these calculations were made with software provided by the vendor. So DPS companies distinguished themselves, in large part, through their software, sold as a service.
Demand for timesharing, on the other hand, was driven by access to generic computing resources (platform) instead of specific software. The first users of timesharing services were engineers, scientists, and other professionals capable of constructing and executing their own programs.
As the timesharing market matured, its value-proposition evolved. The use of specialized software became one of its customers’ primary requirements; though platform access remained valuable. The maturation of the timesharing sector was accompanied by a shift in market level. While enterprises had composed the vast majority of timesharing customers, the SMB market was increasingly perceived as a legitimate and valuable target.
The Character of the Market
The timesharing market’s reach was limited to the “already computing” market segment. More than two thirds of the companies that consumed timesharing services owned their own computer hardware. The same is true for today’s US/European enterprise SaaS adoptees. The lagging of the overall enterprise SaaS adoption rate (relative to SMB) has historically been overstated ; but barriers to adoption are higher for enterprises with existing infrastructure than those without.
SaaS installations in the US and Europe more often replace installed, legacy systems than those in the Asia-Pacific region, which are more often new deployments. The prevalence of legacy software installations exerts pressure on the US and European SaaS markets to provide solutions tailored to enterprises with significant IT infrastructure investment. Two complementary requirements of such enterprises are flexibility and simplicity: they demand SaaS which interfaces with existing infrastructure; and their SaaS investments are supposed to reduce maintenance and/or configuration overhead.
Service oriented architectures (SOA) provide sophisticated interaction with a software product’s components via well-defined interfaces. These interfaces provide customers the tools to customize and automate interaction with existing infrastructure. Customer-facing SOAs widen the market envelope to include customers with complex IT infrastructure, while leveraging those customers’ IT personnel.
In general, SaaS benefits from its automatic, universal software delivery mechanics. But hybrid SaaS deployments must strive to avoid version skew among its on-premises components. Version skew is a major contributor to engineering technical debt. And importantly, non-uniform experiences among customers may complicate a hybrid SaaS product’s market-perception.
The timesharing market grew exponentially until computers became small and cheap enough that businesses preferred their purchase and maintenance to the consumption of cycles and storage on a remote mainframe. Then the DPS market contracted quickly, disappearing in less than a decade. Such points of inflection are not necessarily unpredictable. After all, the price and size of computers had been decreasing for decades prior to the timesharing market’s contraction.
The PC revolution’s preemption of SMB timesharing adoption doesn’t negate the vitality of its market-growth prior to decline. In 1965, the entire DPS market was one fifth the size of the systems market in terms of revenue. By 1978, the proportion was approximately one half. Timesharing’s share of the DPS market increased approximately 50% within the same period.
Before PCs flooded the market, technology pundits imagined the future of software delivery as similar to a public utility, an age of universal access to powerful, remote computing resources. That antediluvian dream seems to have reemerged, tempered by history, in today’s market, which is saturated by IT infrastructure. Perhaps tomorrow’s infrastructure will be neither on-premises nor remote, but distributed.
1) Benlian A. and Hess T., Buxmann P.  Drivers of SaaS Adoption — An Empirical Study of Different Application Types. Business & Information Systems Engineering, 5:357-69.
2) Campbell-Kelly M. and Garcia-Swartz D.  Economic Perspectives on the History of the Computer Time-Sharing Industry. IEEE Annals of the History of Computing, 30:16-36.
By Dane Van Dyck, Axcient Engineer