Long transmission delays such as those inherent in satellite
communication are most certainly a cause for concern among users of
remote interactive systems. Since the University of Hawaii will, by
the end of this year, be linked to the ARPANET via satellite, the
consequences of such delays are perhaps of more immediate concern to
us than to current members of the surface net. Consequently the BCC
500 research group here has been studying various solutions to the
problems of buffer allocation, partitioned echoing, etc. re-introduced
in RFC 346.
Generally, the solutions come from extensions to the original
design concepts of the BCC 500 distributed communication system. The
500 was designed to serve a large number of geographically-scattered
users each of whom accessed the central computing facility through one
of several remote concentrators. [The concept is not too unlike that
of users at different TIPs all accessing a single host.] Since it was
felt that in full-duplex, character-by-character interaction, echo
delays of any noticeable length should not be tolerated, a facility
was provide whereby the concentrator could provide local (to the
terminal) echoing when deemed appropriate. (A character input/output
microprocessor, the CHIO, in implicit conjunction with the terminal
user's process executing in the CPU dictated when it was appropriate.)
The problems associated with coordinating the concentrator and CHIO in
the partioning of echoing were solved for the BCC 500, but are not
immediately extensible to the asynchronous message transmissions of
the ARPANET - especially with the introduction of satellite delays.
As stated, we are working on some viable alternatives.
It is not known, at present, what effects the incorporation of
these partitioned echoing techniques might have on the existing net.
Perhaps local echoing will become a function of User TELNETs; most
certainly local echoing should be available in the TIP. But could it
be incorporated into the IMP so that TIP and User TELNETs can be used
without change? If so, what happens to the concentrator's local
echoing capability in a system such as the BCC 500?
These questions do not have immediate answers. Other problems
obviously exist because of the differences in serving system
conventions for terminal control. We, in conjunction with the ILLIAC
group at NASA-AMES, are seeking solutions to such problems in general,
with an eye toward their implementation