The inter-nodes communication bottleneck has been a key factor that restricts the large-scale expanding of scalable router's software architecture.To solve the problem,this paper introduces a transmission adapted sub-layer in the supporting model of traditional software architecture.Through feature extracting of the up-going data stream and pattern matching with registered task,data stream can then be classified and divided based on the content of information to increase the effective communication rate.The paper further analyzes the model's performance from task's three characteristics:Distribution rate,spread number and traffic rate,and provides an optimized task dispatching reference.It shows that the introduced transmission adapted sub-layer can reduce inter-layer redundant flow and extensibility bottleneck of communication.Finally,the presented experiment verifies the theoretical analysis.