Network Working Group J. Melvin
Request for Comments:
153 R. Watson
NIC:
6758 SRI-ARC
15 May 1971
SRI ARC-NIC Status
Computer and Network Status
The conversion to the DEC PDP 10, running the BBN operating system
Tenex, has just about been completed. We have had a number of
obscure bugs which caused delays recently. Several symptoms were
traced to bad data being written into memory. This problem was
diagnosed as a noisey ground on a chip in the drum-disk memory bus
access control. With the problem fixed our reliability has improved
significantly to about one crash every day or two. System attention
has now been turned to system measurement and tuning and to bringing
up an NCP and Telnet.
We have been working to bring up the BBN NCP of Doc. #1 NIC (5143,)
and BBN's Telnet. Because of our different configuration from BBN's
and slightly different system we have not yet removed all the bugs
caused by these differences. As of May 14 we estimate that we are
only a few hours away from completing this task. We need more
testing before we can provide network service. We will bring up the
NCP of
RFC 107 NIC (5806) when we can obtain it from BBN and the
official Telnet when it is specified and BBN can provide it to us.
At present our local connect capacity allows for 12 displays and 24
typewriter terminals. With about 10 displays and 6 typewriter
terminals running NLS, response is satisfactory, but marginal for
display users. The delivery in June of new Bryant drums and the
measurement and tuning in progress should increase capacity and
response. How much improvement to expect is not known.
The system processing required to support a network user is heavier
than that required to support a local typewriter user. Therefore we
are not sure how many network users we will be able to support
without degrading response seriously or requiring us to limit local
loading by administrative restrictions. Our guess at the moment is
that we can handle 6 network users by middle summer with an
optimistic expectation that we might be able to handle closer to 12.
As there is only limited interactive experience over the network, we
do not know what its response characteristics will be like. We may
find that the delays caused by two timesharing systems and the
network transmission may allow us to support the higher number of