From : asli+1oLfd13U
"
The tragedy is that upper management very much wanted new revenue streams. We tried to enter Retail, Financial, Education, Computer Security, Energy, and other markets — even Video Games. Yet with few exceptions, the people chosen to lead these efforts did not have the skills to succeed.
"
In the case of several of these product verticals, there was an even bigger problem -- one that ultimately impacted SAS' ability to grow its core analytics market share as well...
Lack of a truly cogent, comprehensive and flexible platform architecture for the Internet and cloud age -- after MVA had reached its growth potential in the prior paradigm of multi-vendor hosting on local, predominantly single machine customer installations.
Enter the 2000s where the major managerial players in R&D often did not play well together, lacked the collective insight/talent to take on this mammoth task (one that know single individual could've possibly succeeded as chief architect at) or were simply too busy, polishing the horns on their division/products.
This timing coincided perfectly with the rise of product management and the corresponding decline of most individual contributors doing any significant basic research to advance the SAS core architecture.
Consequently, a mishmash of fifedom developed components like SPDE, TKTS, (and no doubt others others that I cannot quickly recall) attempted to fill in gaps while MVA was extended with TK (Threaded Kernel) in order to better exploit multicore/multithreaded host capabilities.
Then the in-database initiatives, LASR, the SAS Metadata Server and ultimately a Java-based, mid tier and web stack came into the picture. These things helped create piecemeal new features and capabilities needed by existing and a relatively handful of new customers.
However, as cool, as some of these features were, they did not add up to an architectural evolution. It would enable the kind of flexibility and power necessary to serve as the underlying computing base for products that could generate _significant_ new revenue.
Hence, a lot of the product teams wound up, building their own platform capabilities, including custom access control subsystems (rather than using the Metadata servers generalized API for such), using external open source, databases, etc. all because the SAS architectural platform failed to provide the core set of services they needed.
Also, consider that these custom implementations created compatibility issues between products, required custom documentation and specialized tech support. This is highly inefficient and antithetical to the good design.