On Thu, Jun 26, 2003 at 11:48:44AM -0700, Rick Warner wrote: > On Thu, 2003-06-26 at 11:40, Javier Gostling wrote: > > > > > It will depend on the specific situation. Compression will do lots of > > good for bandwidth scarce situations, but on a LAN or standalone system > > it will just waste CPU. > > This is so lame. Any PC less than 2-3 years old and not being used > as a server (which should not be running X anyway) has so many spare CPU > cycles that the amount taken to compress the stream will be trivial. > With today's CPUs there is no valid argument against compression.
My point precisely. I have a 5 years old P2/300 at home running RHL9. It sometimes runs 2 X sessions concurrently, one local, one over a LAN. This is in addition to some server stuff (mail, proxy, etc.). In this scenario, CPU usage IS a concern, and compression would most likely reduce performance instead of increasing it. Another issue (derived from the dual X sessions above) is scalability. How scalable is a compressing protocol? What would be the consequences of compressing data streams in a 50 user multiuser application server? My instincts tell me it would be disastrous. I agree that for a single user workstation on a modern CPU it would make a definite improvement, but my point remains. If this was implemented, it should be an option so as to have the choice to use it where it will improve performance and dump it where it will not. Cheers, -- Javier Gostling D. <[EMAIL PROTECTED]> -- redhat-list mailing list unsubscribe mailto:[EMAIL PROTECTED] https://www.redhat.com/mailman/listinfo/redhat-list
