EaStMAN MOC: Proposed Private ATM Address Scheme

Proposal for EaStMAN Private ATM Address Structure

A scheme for private ATM address scheme within EaStMAN is proposed. It follows the recommendations made in the UKERNA interim private ATM addressing scheme.

The major features of the plan here are

to provide a topological addressing scheme for the EaStMAN partners to define their ATM addresses;
to be focussed around the ATM ring as a feature of the network topology;
to provide suggestions for on-campus addresses.

To aid in following the assignments, it helps to be familiar with the topology diagram.

The Scheme

The proposed 13 octet prefix for EaStMAN on on campus entities is (dots are employed to separate fields):

            39.826f.1107.16.7000.00.nnmm.ee.ff

where

```
  39    AFI - ISO DCC
```

  826f  IDI indicating UK (with padding 0xf)

  1107  CFI/CDI (country format and domain identifier).  These
        BSI defined fields identify UKERNA as provider

  16    Region/topological octet: indicates attachment to
        SuperJANET at EUCS

```
  7000  EaStMAN site code
```
```
  00    UKERNA placeholder (must be 00)
```
```
  nnmm  Institution and EaStMAN `ring access point'
```
We considered that it may be useful to have a fairly short mask by which we could identify the owning institution for the address. Thus these two octets are a combined field of institution - `nn' - and ring access point - `mm'. Note that the access point is notional: Stirling have no direct attachment to the ring but are assigned a value as if they were.
Enumerating the ring access points is a fairly stable operation if we include the FDDI-only sites, since it is unlikely that we will expand that core in the near future. Spurs would seem to be the way we will enlarge the MAN system.
The values for the institution are:
```
      01        University of Edinburgh
      02        Moray House
      03        Queen Maraget College
      04        Napier University
      05        Heriot-Watt University
      06        Edinburgh College of Art
      07        University of Stirling
```
Including the placeholder values for FDDI-only sites, the values for the various ATM switches are (assigned anticlockwise from King's Buildings):
```
            
      KB        0101
      Pol       0102   - placeholder
      OC        0103
      MH-H      0204   - placeholder
      QMC-L     0305
      NewC      0106   - placeholder
      Nap-M     0407
      H-W       0508
      MH-C      0209   - placeholder
      QMC-C     030a
      Nap-S     040b   - placeholder
      ECA-L     060c   - placeholder
      ECA-G     060d   - placeholder
      Stir      070e
```
The next two octets. ee ff, are free for assignment by each institution. As an example, University of Edinburgh are trying out a scheme where:

  ee    Campus number available off the man switch:

      00    Attached to EaStMAN ring switch
      01    Western General Hospital
      02    KB
      03    Appleton Tower
      04    etc.. see George

  ff    Switch number under the campus octet.  (Might want 
        this the same as any SPANS (unique on your net) area number 
        and/or drop ee significance).

Alternatives

It is possible for the HEIs to use their own site codes instead of the EaStMAN code - this would allow recognition of the HEI very quickly from the form of the address - and one would not need an artificial construct as per the above scheme. It would also imply that the mask which defined a point of access on the ATM ring could be shorter - it would include the HEI site code and at least one other octet to enumerate the switch. Thus, we would save probably 1 octet - or rather make one more octet free for assignment. In addition, although we might see the institution address separated by another (e.g. QMC-L between UoE OC and Nap-M) the routing would probably not be difficult to configure (as also PNNI). However, it would be more difficult to create a mask `all EaStMAN' which might be desirable in the future (anycast services) - this would need to reply on the region octet which might not be a one-to-one map with EaStMAN in the future.

EaStMAN partners are invited to consider and comment on their preference as soon as possible.