This brings up the question of thread-safe coding, as was discussed in the thread "a random question" by Alan Wolfe on 9/24/02.
It seems to me that this threading model now imposes new requirements on applications programmers to assure that their shared variables are properly protected. It seems to me that writeable variables in the application and server scopes now need to be wrapped with protection since they can be changed by other threads. >Reproduced from http://www.stonesteps.ca/faq.html > > >Q:What is preemptive threading model and how will it affect the >performance?� > >A:All incoming application requests are handled in the application server by >worker threads, which receive requests from the Witango client, retrieve the >requested application file, process queries associated with the application >actions, generate the result HTML out of actions' HTML fragments and send >the results back to the client. The way threads interactbetween each other, >the server core and the operating system is called threading model. There >are two basic kinds of threading models: cooperative and preemptive.� > >In cooperative threading model, only one thread is active at any given time. >After working for some short period of time (typically 300 ms), the active >thread voluntarily pauses and allows another worker thread to process >another request. In preemptive threading model, the operating system >controls the scheduling of all worker threads and allows each one of them to >become active for even a shorter period of time (typically 30-60 ms), after >which another thread is made active and so on. The cooperative threading >model has many inherent disadvantages: synchronization delay when threads >have to be interfaced with the preemptive outside world (e.g. DBMS servers), >inability to take advantage of multiple processors (CPUs) and susceptibility >to the potential danger of the active thread not yielding properly and >therefore blocking other worker threads and their requests. > >Tango 2000 uses mixed threading model with the server core and all worker >threads running in the cooperative mode and some of the external operations, >such as launching command line applications of an External Action, executed >preemptively. Every time a preemptive task has finished, the cooperative >worker thread that launched it has to synchronize with the rest of the >worker threads before it can continue its life cycle. > >The Witango server is a fully preemptive server and is free of the >disadvantages of cooperative threading described above, which means that >those application files that interact with other services (i.e. running >command line applications, SQL queries, etc), will run faster as long as the >external services do not share the CPU with the application server (e.g. >DBMS is located on another machine). For example, an application file that >reads/writes large files with a File Action will run faster on a preemptive >application server because the disk controller uses Direct Memory Access >(DMA) mechanism to transfer disk data into memory and while one worker >thread is waiting for the File Action to complete, all other worker threads >are processing other queries.� > > > >On 12/11/02 10:05 AM, "Chuck Lockwood" <[EMAIL PROTECTED]> wrote: > >> The threading model has been updated in v5. "Witango App Server5 now runs in >> full pre-emptive multitasking mode." >> >> What does this mean exactly, in processing terms? v4 processes multiple >> requests and seems to interact with external systems (databases, mail >> servers) concurrently. >> >> So what's the big deal? >> >> Chuck Lockwood >> ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ >> LockData Technologies, Inc. >> 309 Main Avenue, Hawley, Pa 18428 >> 570-226-7340 ~ Fax: 570-226-7341 >> [EMAIL PROTECTED] ~ www.lockdata.com >> ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ >> >> >> >> ________________________________________________________________________ >> TO UNSUBSCRIBE: send a plain text/US ASCII email to [EMAIL PROTECTED] >> with unsubscribe witango-talk in the message body > >________________________________________________________________________ >TO UNSUBSCRIBE: send a plain text/US ASCII email to [EMAIL PROTECTED] > with unsubscribe witango-talk in the message body > Bill Conlon To the Point 345 California Avenue Suite 2 Palo Alto, CA 94306 office: 650.327.2175 fax: 650.329.8335 mobile: 650.906.9929 e-mail: mailto:bill@;tothept.com web: http://www.tothept.com ________________________________________________________________________ TO UNSUBSCRIBE: send a plain text/US ASCII email to [EMAIL PROTECTED] with unsubscribe witango-talk in the message body
