چكيده به لاتين
Due to the growth of geographic scale, the number of users and variety of services, communications and traffic volume has increased in the computer networks. So, according to requirements of the networks, new features such as multipath architectures and distributed resources have been substitute with the architecture of traditional networks. In software defined networks, with the separation of the control plane from the data plane, the network can be managed more efficiently and more programmable. So, unlike traditional network management approaches, software defined networks are appropriate for Non-traditional networks management issues, such as the distribution of load between resources. Despite the existing methods of balancing the load between servers, these methods don’t consider the path between the client and the server. In addition, these methods are not sufficiently dynamic to match the different conditions of the network.
To address these challenges, in this research we provide a dynamic, path-aware method to the distribution of loads between servers which, besides the load of the server, it also consider the status of the path between client and server in process of selection. The proposed method attempts to reduce the number of packet loss and latency with making appropriate decisions. Additionally, in this research, control of congestion and overload of resources is addressed by providing a monitoring algorithm. For this purpose, the load of the servers and path to them are monitored periodically, and according to the proposed solution, if necessary, a decision is made.
The proposed method implemented by Containernet emulator and Floodlight Controller ,also has been compared with the server selection algorithm for different purposes (a method in which the path, is not involved in the initial selection of the server). Evaluations indicate that the number of packets lost is about 90% reduced, throughput is about 12% increased, jitter is reduced to 84% and response time is also reduced by 16%.
Keywords: software defined networking, load balancing, server, multipath, monitoring.
