THINK Home Digital Archive Search Feedback

 

THINK  NPL

A Technical History of National Physical Laboratories (NPL) Network Architecture - A Technical Tour



Hardware

Interface Unit

The interface computer and the node computer can be the same computer. It is the single point through which all network traffic must enter. Access to the network and accounting information for the users is controlled in the interface computer. The interface computer is also responsible for reconciling protocol of the high level networks for the various peripherals because they are all attached to the network through it.(Davies). This is the point where the users connect to the network.

The interface computer is responsible for assembling data in to the fixed packet formats. Additional formatting information such as the source and destination information is included into the packet. The interface computer's control protocol can include serial numbers for the packets in a fixed position of the message area for reordering. Multiplexing and demultiplexing also takes place at this point of the network. Although the network itself provided methods for error control and flow control, the interface computer also proves some form of these services.

The interface computer was originally designed to be implemented using a Plessey XL12 computer. The computer was chosen because of it's "unique multiplexing arrangement" that allowed a large number of devices to be connected. Although this scheme can included a "tree of demand sorters which could be infinitely deep," the NPL project proposed a depth of 3 which allowed at most 512 device to be connected. But in late 1968, Plessey announced that the XL12 will be cancelled and the group decided to replace it with a Honeywell DDP-516. This implied that the responsibility of the demand multiplexer was left to the group, mainly Barlett and Scantlebury who were in charged of the interface computers. (Campbell-Kelly p231)

Interface computer


Node Computers

In the implemented networks, the Mark I & Mark II Networks, a Honeywell DDP-516 computer is used.

Services provided:

  • Flow control
  • Error detection
  • Error recovery
  • Routing
  • Congestion control

 

<< BACK >> NEXT

BACK TO TOP

 

Written by the THINK Protocols team, CS Dept, UT Austin
Please direct comments to Chris Edmondson-Yurkanan.

This document was last modified on Tuesday, 11-Jun-2002 10:18:02 CDT.