@make[report]
@device[x2700]
@flushleft[Edinburgh Regional Computing Centre]
@flushleft[For the attention of the E.C.U.C.]
@majorheading[Report on Communications]
The general level of the network service to users has been significantly 
better this term.
@paragraph[CPSEs]
For the first time ever, the main KB CPSE has run for a period of three
months with no unscheduled hardware or software breaks.  There were two
failures in the period caused by mains power outages.  Over the three
months the CPSE was taken out 4 times at lunchtime, twice to fix simple
line faults and twice to upgrade the software.

The second KB CPSE had two software crashes on the 14th march, otherwise no
faults in the 3 month period.  It has been taken briefly out of service in
the last few days to change network routing otherwise it has not been booked
out.

The Appleton Tower CPSE has not performed as well as the other two.  There
have been 8 hardware crashes and two software crashes over the period.
It has also been taken out at lunchtime 8 times to allow GEC to investigate
problems.  The continuing problems with this machine has prompted the action
which is outlined below.
@paragraph[Future Plans]
A new  switch, called a @B{TelePAC}, made by Telefile,
 has now been used in trial service
at KB since January 1986.  We have now been awarded money from the Computer
Board to buy a 30 port version of this switch as a pilot for the Academic
Community as a whole.  This 30 port
switch has been installed and the load is being
slowly built up on it.  Its introduction is being used as an opportunity to
increase the overall resilience and performance of the network, particularly
to the Central Area users.  The smaller GEC switch at King's Buildings, which 
has proved highly reliable in the past, is being moved down to the Appleton
Tower to provide an alternative backup to the existing 4190.

The 4 switches in EDNET are  to be fully interlinked  using 4 Kilostream
circuits running at 48 kb/s, these circuits have been installed and we
are  about to start testing them out.  A diagram of the structure of the
changed network nodes is below.  There will be spare ports on the switch in
the Appleton Tower and a procedure for switching lines will be
established.  For critical users,
 such as a student class, the mainframe services
will still be able to be provided even with the loss of any one of the four
switches.  Multi-user minis which are only connected to a single switch will
still be at risk but alternative patching arrangements will need to be established
for those whose service is vital.
@paragraph[PADs]
Although a few PADs, notably KB4, have caused problems, generally the performance
has been reasonable.  Mains problems have been seen and PADs which have
crashed have been fitted with suppressors which have significantly reduced
the failure rate.  Each PAD is now polled every night 
to gather statistics
so that a comprehensive
picture of the level of service provided can be built up.
@paragraph[TCPs, FEPs, PSSE, XCALL]
The rest of the network components have been performing satisfactorily.
@flushright[B. A. C. Gilmore
12th May, 1986.]