y “MANAGED>AMAREL NOTES} MON SeMAYe7S 6f43PM lo 5 Anak?

Saul! e= Tom and I have given alot of thought to ways (nm which a Pigoroys
prepasel for augmenting the Rutgers facility eauld be formulated on the
rationale of further service to SUMEX@AIM as a community: end distingusihed
fram what would be the selfwavident merits of further SYPPOrE to your awn
research project, [The jattere IT believes would have to be given specific
review by ean IRG im terms of a Separate specific proposal) and I think we
must be very careful =we.g, 19 anticipation of eventual competitive renewal
“= not to (nvade the prerogatives of that preview procedure under the colora=
tiom of a BUMEX@AIM facilitation, And you will mote this {8s a private
message (nf the light of the abvious delicacy of its substance,)

We certainty ean see some benefits along the limes ef the backyp
expedient we had discussed; but there are a lot of problems and we wonder
if the full expense, to US as wel} as to YOU, of staffing so as to maintain
day*torday plugetoeplug compatibility, as well as of the hardware costs,
18 Mot too much for the anticipated benefits, We have so many problems al=
Peady in such areas a8 LISP version maintenance in maintaining program come
patibility that we foresee a very substantial burden if we Jointly undertake
the obligetion of deing this im response to such diverse user groups as
wauld be imvelved im Rutgers cum SUMEX AIM, Nor would you want te delegate
the mMamagement decisions ta SUMEX staf? that would be the most efficient way
to accomplish this geal,

Perhaps this {8 sti)] mot impossibiey it ie after all pe(marijy in
the backup mode that the PERFECT compatibility weuld have to be implemented)
so that 4¢# Your charter assured that, on Jegitimate eal! from SUMEX#AIM
(the detailed rules fer which would be in the hands of AIMEXEC, but such
as te assure that vital demonatrations, clinica! applications, report
generation and the liker gould be continued at Rutgers during scheduled
SUMEX downtimes and im some measure im response to unseheduled), that
Rutgers would mount the CURRENT SUMEX vermion of TENEX and of support
4iJee Clike the FORTRAN, SAIL and LISP. libraries), some more comfortable
latitude might be feasible for speciajization of the two systems, Unfors
tunately, some of these systems are so written that users do not have the
option af calling on one or another alternatives and the software effort
te remedy that stupidity would at this point be prohibitive,

Rainer Schulz will shertiy message you about some measures of the
effort that would be (nvelved im your bringing ue TENEX, That should not
be dif#{ecult, You will else reautre probable one fulltime additional pers
Som with the resonsibility of Jfaisonm with SUMEX, and we may indeed have
te augment our own setAff almost as much for the sAme purpose,

And then we have te solve the problem of file transfers DATACOMPUTER
18a Possible future option, (@ (8 down at thie very moment, and it is
obvious that it is mot a reliable expedient this year, In all candor, I do
believe that from the SUMEXAIM perspective, the proposed investment
moe and the probable future costs that wil) follow with {t == are the most
cost effective use of available funds, [For example, im a year op twos when
we are more heavily loaded, the upgrading of the KI to a KL processor ean
prebebly give us # 2 ef 3X increase im thruput at relatively little cost, We
are at a point on the investment=yield curve (taking stafing also inte
@ccOUnt) where (t makes Jittle sense to diateibute the investment into frag=
Ments,

HOWEVERs there {8 another area where {t would be sensible to experime:
with some diversification of service, and the augmentation of your facility
y @MANAGEDAMAREL NOTES) MON SeMAY@7S 6543PM

might be a very effective way to ga into this, namely ¢f you were to fecus

on a totally different mode of COMMUATEY SUPPOPE, OsGeer bY FURMing TOPS#10

for its monwTENEX gyecessors) for the benefit of SUMEXPAIM Users who are havil
di¢¢iculties in adapting their programs to TENEX! We believe many of these
Problems will eventually be solved, but some perhaps only at umpeasonable ex
pense compared to just having time available on a TOPS@{0 machine, Also the
conversions would be very much facilitated i* we could examine the running
code as it was IN FACT being executed wader TOPS#10,

In our opinien, this widening of options to SUMEX@AIM users would be
of greater benefit as a return for augmenting your system, than would be
the exact duplication of SUMEX#tanmex, With efficient commyunieation between
our two systems, there may even be occas{ons when ysers would run program
Segments alternately under TENEX and under TOPSe10 {nm accordance with the
availability of the appropriate software,