Layering in Glass Net

I have claimed that the ordering of the layers of functionality largely determines architecture. I explore that claim in detail here regarding the Glass Net scheme. (It is probably relevant that these ideas were meant to support communications applications and that the nodes were oriented towards communications or were closely associated with such comm gear.)

First we assume nodes, which are a somewhat homogeneous collection capability based software platforms.

Next we assume communication links between some pairs of these nodes. This forms a network and we will think about a connected component of this network. These links are physical and thus a complete cross bar is not suitable. The links are not assumed to be secret but perhaps they might be unspoofable.

Both ends of each link are owned by a link agent which resides in the node at the link’s end. The two link agents at opposite ends of a link are peers and trust each other. If the link is spoofable, then the agents impose a strong message authentication code between them and that seems to require a shared secret. The two agents for a link will use DSR packets and each keep the link accumulator.

The base capability layer will be enhanced to provide a scalar money value of each Ob in the system. While this is not strictly necessary it may be practically necessary for the intended uses.