herzlich@NGP.UTEXAS.EDU (Larry Herzlich) (11/12/87)
I've received a couple of responses regarding my concern for using terminal servers to a MicroVAX 3xxx system. Most of the comments stated that I was way off about the traffic from terminal servers (Jerry Leichter, Joel Snyder) and they may be right. The worst case scenario for terminal traffic is estimated to be around 10% of the bandwidth and that's if everyone was typing as fast as they could at 9600 baud. I guess the real concern is the LAVc traffic and the DEQNA or DELQA. Our planned configuration: 1. VAXserver 3500 with 4 large disks (RA82 or CDC drives) file server 2. 3 - MicroVAX 3500's (using the RA70 for Swap/Page) compute server That gives us 60 potential users plus the traffic of the LAVc System Communication Services (SCS), plus the normal file transfer traffic between our systems and other systems, plus DECnet traffic, plus TCP/IP (WIN/VX) traffic. Finally, during off hours we will do BACKUP to our CI cluster tape drives. This seems like a lot of traffic to start with. My concern is based on the performance session of LAVc from the Nashville DECUS. I honestly don't think that the DEQNA or the DELQA can support all the traffic outlined above. Do you? One interesting comment was made by Nigel Arnot of the UK. He mentioned using 2 Ethernet devices. At 3 * ~$3K per controller, plus the cost of 3* the cost of tranceivers and drop cables, isolating my terminal traffic now costs: $30K for DECserver 500 (This is probably low for 64 lines) $12K for the 2nd controllers connectors, and cables Now, for 64 terminal lines that makes ------ ~$650 per port with 2 Ethernets and ~$470 with one Ethernet I really need a cheaper way to handle terminals. I was hoping there was a DEC or 3rd party solution to getting terminals into the VAX 3xxx boxes. ---- 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