Multipath directing routes use a few ways to disperse activity from a source to destination. This enhances execution as well as accomplishes stack adjusting and adaptation to non-critical failure. Despite the fact multipath directing isn't conveyed generally in the web, current frameworks pick the ways with the equivalent most reduced expense to the goal and the equivalent managerial separation. In this paper, we propose adaptable multipath organize wherein the multipath directing routes calculation is conjured when the nature of administration is influenced. Our proposal gives a technique to move step by step from existing system solid multipath arrange later on. We have utilized machine learning way to deal with distinguish the occasion when arranging stack offsetting is required with significant outcomes.

Keywords: Multipath routing; load balancing; QOS; SLA; Neural Network; Naïve Bayes