White PapersNew programming technologies often go through an early period of upheaval as developer communities begin to separate theory from practice. Initial naivete is tempered by realism and experience. Witness the history of the Java programming language: the initial focus on browsers and not-so-thin-client applications has matured, broadened, and shifted to embrace just about every level of computing - from consumer products to the data center. Multithreading has seen a similar evolution, and one of the shifts in emphasis has been away from “threads for everything” to “threads for scalability.”
For example, developers used to believe that a threads stack was an ideal place to store application session state. However, experience has proved that it is generally better to hold session state in well-designed data structures, and deploy a worker thread pool to deliver scalable, predictable performance.
