Research On Fork-Join Scheduling Strategy For SDN Multistreaming

Software-defined network(SDN)is a new network architecture,which can update the network configuration in real time and accurately,and meet the service requirements of the emerging new data services.However,with the advancement of digitization and the development of the digital economy,the network is carrying more and more services,and SDN is also facing many challenges,among which the scheduling and processing of a large number of business data streams are particularly prominent.Therefore,a new scheduling strategy and processing model are needed.Although some progress has been made in the research on data flow scheduling and processing in the industry,relevant studies have not fully taken into account the network architecture characteristics of SDN,which are shown as follows: 1)The processing of data flow needs to fully consider the flexibility and intelligence of network resource scheduling,so as to meet the requirements of existing data services on SDN network architecture;2)The collection flow needs to be further divided into multi-class data flows.A single scheduling algorithm cannot fully guarantee the fine granularity and the service requirements of all data flows;3)The data flow scheduling strategy in SDN is configured by the controller,and the existence of the controller should be taken into consideration when studying the data flow scheduling and processing.Therefore,this paper studies the scheduling strategy for SDN multi-data flow introduces the fork-join model and USES its parallel processing capability to provide faster services for a variety of data services.The main work is as follows:(1)Based on the current network data flow variety and differentiation characteristics of service demand,the introduction of the Fork-Join model to flexible allocation of network resources,use of the advantages of the Fork-Join system parallel processing data,reduce the data stream processing delay and backlog,etc.,and using the network routing and queuing theory analysis to compare the Fork-Join the backlog of upper bound and time delay.(2)The multi-data flow fork-join scheduling system model(MSFJ)is established based on the priority partitioning of different data flow businesses.Meanwhile,a multipriority queue scheduling scheme is proposed based on the combination of multiple data flow scheduling strategies.Moreover,the overstock and delay upper bound of MSFJ system under different scheduling schemes are derived by using the network calculus theory.(3)Numerical simulation was used to compare the backlog and delay of MSFJ system under different scheduling policies.Meanwhile,a data flow platform of SDN was built based on MSFJ system model to further verify the advantages and disadvantages of different scheduling policies and provide an experimental basis for multi-data flow scheduling.To sum up,this paper conducts an in-depth study on the existing scheduling strategies and establishes a scheduling model for multiple data streams in SDN.On this basis,it proposes a multi-priority scheduling strategy to reasonably allocate network resources to meet the network requirements of various data services.The network algorithm theory is introduced into the performance analysis of multi-priority scheduling strategy,and the fork-join model is introduced to improve the system's processing capacity,so as to meet the scheduling requirements of multi-priority data services.