@h[Phasing Out the Ring and Conversion to XXX]


@s[Phase 1. Creation of Additional X25 Ports]

An additional Camtec PAD (with an unmodified back plane) will be
connected to the @b[KB CPSE] and used as a switch.  In the first
instance at least four existing machines could be attached to this
PAD as the combined traffic is low.

          1. SIAE   RSX11  (line 30)
          2. Physics RSX11 (line 16)
          3. PRC Prime     (line 35)
          4. MRC VAX

It is important that they appear as the @u[same] network addresses
when connected to the PAD to avoid large changes to address tables.

          @u[Action:] Duncan Walker, Jim Leitch,
                  Callum Young, Derek Clyde.

          @u[Timing:] Experiments should start as 
                  soon as possible.

@b[Note:] The remaining asynchronous ports could be used as @b['normal']
PAD ports.

@s[Phase 2 Transfer of Ring TCPs to X25]

A DUP11 (or equivalent) needs to be fitted to each of the TCPs and the
TCPs moved onto the freed ports on KB CPSE.  New addresses will need
to be used - 15000090-15000093.

It is not desirable to run with 48 ports for any longer than absolutely
necessary and steps should be taken to move some of the users from these
TCPs onto PADs.  It should be possible to move the EMAS team soon as 
they ought to be on PADs to facilitate access to the new machine.

          @u[Action:] Jim Leitch, Peter Macvean, Brian Gilmore
                  Derek Clyde.

          @u[Timing:] The first one could be taken immediately,
                  the rest after Phase 1 is complete.  The
                  entire shift should be completed by the
                  beginning of the term.

@s[Phase 3. Transfer of Ring FEP to X25]

The Ring FEP will be freed of interactive traffic after the end of Phase
2.  It will then be available to take some of the XXX load from the
existing FEPs as the number of PADs on EDNET is steadily increased.

Filestore traffic will need to be routed through XGate wich ought to be
able to handle it as it will only be handling interactive traffic to
FSTORE and PSSE and PAD to Info traffic.

FTP traffic to PSSE will need to be routed via Rutherford.  Slow Devices
will need to be moved to X25.

After evaluation of the relative loads, a DQS11E could be moved from
XGate to FE4.

The second line on FE1 at BUSH (currently unused) should be attached to
EDNET to take some of the increasing PAD traffic.

          @u[Action:] John Henshall, Brian Gilmore, 
                  John livingstone.

          @u[Timing:] 1st or 2nd quarter of 1985.

@s[Phase 4. Conversion from ITP to XXX]

A gradual switch from the use of TCPs to PADs will be made during the
entire period.  It is essential that most users of the new mainframe
are on PAD ports if severe congestion of XCALL is to be avoided.  This
means that several conversions will have to be made during the summer
vacation of 1985.  With the decrease on the 2900s it should be possible
to support the bulk of the interactive traffic using the XXX protocal.

          @u[Action:] Derek Clyde, Brian Gilmore

          @u[Timing:] Throughout.

@s[Phase 5. PSSE]

PSSE should be moved onto the X25 network byf making use of the code
developed at Kent.  This will mean dropping support of ITP on PSSE and
TCPs will need to go through XCALL to use PSSE.

          @u[Action:] Ruth Holland.

          @u[Timing:] During 1985.

@s[6. INFO]

XXX will need to be supported by INFO, this will mean that ITP support
is dropped.  In addition the host polling routines will need to be
changed to XXX for the new mainframe.

          @u[Action:] Scott Currie.

          @u[Timing:] by October 1985 at the latest.

@s[7. Filestore]

It will become increasingly difficult to maintain effective use of the
Filestore as it is isolated on the Ring and the gradual conversion to
XXX protocols.


 
                                                        Brian Gilmore
                                                      6th September 1985.