herzlich@NGP.UTEXAS.EDU (Larry Herzlich) (11/08/87)
In Steve Ward's article about `local area vaxcluster performance', he mentioned using direct lines to the MicroVAXes versus using terminal servers. >Terminal servers can be used instead of hardwired serial ports, but >each multiuser node should have extra memory and derated by a few >concurrent logins to compensate for the extra terminal server overhead. >Steven Ward >Smithsonian Astrophysical Observatory I would like to configure a system with direct lines but I don't see any reasonable way to do this using DEC hardware. Putting 3 DHQ-11's inside a MicroVAX seems like the limit due to panel inserts. I'm not even sure this is possible if we are using a KDA50 controller on the MicroVAX at the same time. I think I'd run out of slots and panels in a normal BA123 box. Now, picture the scenario of the MicroVAX 3500/3600. I can license the system for 20 users or as many as 40 users on the big VAXserver 3600/TU81/RA82 configuration. How the heck can I get 40 users into that system? I believe that the performance is there at 3 MIPS but I don't see any way of getting 40 lines into the H9642 cabinet. Maybe one of the field test sites can discuss the layout of the I/O distribution panel on the MicroVAX 3500 BA213 or the MicroVAX 3600 H9642 cabinet? Remember, I DO NOT want to use terminal servers. I don't like the idea of spending 30K for a DECserver 500. I also don't want to discover that my Ethernet (even with a bridge to isolate the traffic) is going to be saturated by my LAVc terminal traffic. DEC, are you listening? Does `DEC Have it now?' ---- Larry Herzlich herzlich@ngp.cc.utexas.edu University of Texas at Austin CCCS001@UTADNX -- BITnet Computation Center UTSPAN::UTAIVC::CCCS001 (SPAN) User Services Division ..!seismo!ut-sally!ut-ngp!herzlich