It will seem strange to many that DSR suggests that the end user deal in routing issues.
Here is why I think that it is a good idea.
First it is an appointed agent of the user, not the user, that copes with the admitted complexity of routing.
The immediately obvious alternative to routing in the end user’s machine is an agent owned and controlled by the end-user’s service provider.
The contract between the provider and user is that the provider will choose a suitable route.
There are two immediate pit-falls here however:
- This leaves the link between the provider and user out of the equation.
This may well be the weakest link and in most need of prioritizing.
It would be well to cover the last link like the rest.
There may well be no alternative to routing over the last link but priority on that link can be a significant issue.
- QoS, (Quality of Service) is a many splendored thing.
Some packets can tolerate loss but not delay, some are the opposite.
Not all QoS parameters have been invented yet.
New data delivery contracts will arise and impact the logic of routing.
The end user has the proper incentive to upgrade to whatever agent and routing protocols he chooses.
Integrated service that routes and delivers may well be provided within DSR.
User routing from the start reserves the option for the advantages described above.
User routing also makes the last link less special.
This does not mean that the routing agent must run on the user’s machine.
To bootstrap routing a DSP (DSR service provider) will provide routes to a handful of popular services.