Load-sensitive routing algorithms have been integrated into the Internet routing algorithms since the early days of ARPANET. The drawback of this condition leads to overreaction to changing network conditions and permanent oscillation effects finally caused the idea of load-adaptive routing to be dropped for some decades.
There is a tremendous growth of interest in traffic engineering and in load-sensitive routing in general from the last year onwards.
Many studies testify that with the stability mechanisms integrated these kinds of algorithms strongly outperform the traditional shortest-path IP routing.
In the study of load-sensitive routing algorithms several emerging approaches to load-sensitive routing are presented and the game-theoretic approach of “selfish” routing is explained in greater detail.
Its potential gains and penalties are theoretically analysed. Subsequently simulations are conducted on a network simulator with a routing protocol of this type and several other routing protocols for comparison.
Let us see the difference between load sensitive and load insensitive routing.
Most routing protocols will not pick the best path based on congestion, as it can lead to oscillations and instability. That is, the link metrics do not depend on current load level.
Consider an example, if some link is loaded and its metric reduces, then all flows may move away from it, which leads to load on another link, and all flows will move back again.
The advantages of the load-sensitive routing algorithm are as follows −
The ability of dynamic routing to congest links and improve application performance makes it a valuable traffic engineering tool.
Therefore, deployment of load-sensitive routing is hampered by the overheads imposed by link-state update propagation, path selection, and signaling.
There are some problems with a load-sensitive routing protocol, such as −
Higher overhead on routers and especially instability.
For example, reacting to out- of-date information or interactions with other self-regulatory mechanisms such as TCP's congestion control.