To: Marvin Mansky who wrote (19462 ) 9/8/1999 10:28:00 AM From: QwikSand Read Replies (1) | Respond to of 64865
Marvin, the details will have to come out in the announcement, but it appears that the main difference is the difference between a smart terminal and a dumb terminal. On the old Javastations, when the user wanted to run a program, the progression went something like this: a) Download Java byte code into Javastation local memory b) Let puny local microsparc processor execute local JVM, which in turn interprets Java byte codes and displays graphics on local screen c) repeat. The new Sunray is more like this: a) Start program on big hairy powerful server that is dedicated to your personal Sunray station b) When you want to run a program on your Sunray screen, send a couple of keystrokes or mouse clicks to your agent program on the server c) The agent on the server executes the program (Java or otherwise) and sends the output back over the network to your Sunray in the form of a stream of screen graphic data. The Sunray does nothing but put the graphics data up on the screen. There is almost no compute burden on it at all. The good news: your compute work, instead of being done by a local processor is done on a big powerful server. Also, your local station is just a screen; all the administration and maintenance is done by somebody else on the server (to a good first approximation); you don't worry about it. The bad news: if there are enough people using this big powerful server at the same time, it could get very busy and slow down like any other oversubscribed computing resource (including the CPU on your current-day Wintel desktop PC when you're doing a Photoshop transform). So people using SunRays have to make sure they buy plenty of big powerful high-margin Servers from Sun to make their workstations stay fast at all times. (James Nicoll note: The server being a single point of failure is not part of the bad news. Sun offers configurations that avoid that problem.) Regards, --QwikSand