SOA Trader has measured the growth in the number of public Web services and their operations since early 2005 up til now. The measuring has been based on collection and analysis of Web service interface descriptions in WSDL format. While early 2005 about 1000 Web services with altogether 10k operations were spotted, by 2009 these numbers had reached nearly 25k and 260k respectively.

Anyway, though the SaaS market is expected to grow rapidly in following years and Web services are a part of major SaaS solutions, Web service growth is expected to slow down compared to rapid growth in previous years. The number of public Web services is expected to reach next year 30k and would provide ca 300k operations. The main reason for this slowdown is related to the high degree of redundancy in globally provided data and functionality, which are available through Web services. From one hand side this ensures that different sources can be combined together and from other hand side means that healthy level of saturation has been achieved.

Patel et al [HPWC07] analysed the size of the deep Web back in 2007 and concluded that there were about 400 000 deep Web databases available then. What can we conclude from this number? What we have seen so far is that a Web service provides usually access to a single deep Web database and thus more than 400k Web services are required to expose the deep Web resources. However, majority of these deep Web resources tend to overlap. Moreover, according to Ventrone and Heiler [VH94] data model redundancy of large federated information systems is about 80%. Our analysis of data models of the collected public Web services shows that this redundancy percentage is even higher, potentially more than 95% in the global scale. According to Moody and Shanks [MS03] that high degree of overlap between different application data models is due to the reason that different project teams have a little overview of related data sources leading to data redundancy and duplicated development effort.

Anyway, due to the high degree of redundancy in the deep Web there is no need to expose all deep Web resources to support development of further SaaS applications. Rather the future SaaS applications are expected to make better use of available data and functionality, which is partly exposed as Web services. The former claim is the main cause for slower growth rate of public Web services compared to previous years.

[MS03] Daniel L. Moody and Graeme G. Shanks. Improving the quality of data models: empirical validation of a quality management framework. Information Systems, 28(6):619-650, 2003.

[VH94] V. Ventrone and S. Heiler. Some advice for dealing with semantic heterogeneity in federated database systems. In Proceedings of the Database Colloquium, San Diego, August 1994, Armed Forces Communications and Electronics Assc. (AFCEA), 1994.

[HPWC07] B. He, M. Patel, Z. Zhang, and K. C.-C. Chang. Accessing the deep Web: A survey. Communications of the ACM, 50(2):94-101 May 2007.