Re: Informix vrs Oracle 7.0.x or 7.1.x

In article <adamg.773156940@infmx>, Adam Green <adamg@informix.com> wrote: > >Also, Version 7 PDQ will be on a range of platforms by year end. >Also note that our PDQ facility is built into the architecture of the >server, it's not a tacked on, extra cost, option. I agree with Adam's follow-up, but think it is significant to bring out a different angle on the above statement. The fact that PDQ is implemented through threads is an incredible gain. Without a multithreaded server approach, a significant portion of the aggregate CPU bandwidth of the machine (SMP) is spent in Unix process switching at consumer-producer exchange points, IO completion, and so on. The whole question is whether you find it lighter on CPU bandwidth to conduct a thread switch than for the OS to stop a process, save it's state, monkey around in the priority-based scheduling code, then finally run the "runable" process. What if the runable process has gone cache-cold and a cache flush is required so the process can run on a CPU half way across the card cage, on and on... ________________ Kevin A. Closson, Senior Technical Support Engineer voice: (503)578-3072 Database Group Leader, World Wide Tech Svcs FAX: (503)578-5453 Sequent Computer Systems 15450 SW Koll Pkwy (MS WIL1-502) UUCP: ..uunet!sequent!kevinc Beaverton, Oregon 97006-6063 internet: kevinc@sequent.com