Premium
DSSDN: Demand‐supply based load balancing in Software‐Defined Wide‐Area Networks
Author(s) -
Sahoo Kshira Sagar,
Tiwary Mayank,
Sahoo Bibhudatta,
Dash Ratnakar,
Naik Kshirasagar
Publication year - 2018
Publication title -
international journal of network management
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.373
H-Index - 28
eISSN - 1099-1190
pISSN - 1055-7148
DOI - 10.1002/nem.2022
Subject(s) - computer science , openflow , controller (irrigation) , software defined networking , load balancing (electrical power) , traverse , quality of service , computer network , distributed computing , mathematics , geodesy , geography , agronomy , biology , grid , geometry
Summary One of the unexplored research areas in Software Defined Networks (SDN) is load balancing of control messages ( e.g . p a c k e t _ i n ) among distributed controllers in Wide Area Networks. In SDN, on every unsuccessful match in the flow table for the incoming traffic flows, the switch sends p a c k e t _ i n to the controller for further action against the traffic flow. The p a c k e t _ i n messages are one of the major contributors of the control request (load) received by the controller. When it exceeds a certain threshold limit, the response time for the control request increases nonlinearly due to the over CPU utilization and congestion. When the controller gets overloaded, typically the OpenFlow‐enabled Devices (OFDevices) are migrated from the current controller to another under loaded controller domain. This migration might cause large degradation of end users' QoS metrics. To resolve this issue, we introduce basic demand and supply curve based DSSDN, a new load balancing method that utilizes the load factors of Software Defined Wide Area Networks controllers. This method selects the OFDevice which causes maximum load on the controller and traversing minimum users traffic through it. The Karush‐Kuhn‐Tucker conditions are employed during the optimal controller selection by the OFDevices to improve the response time effectively. During implementation, virtual threads running on the controller representing the OFDevices are used to take the optimal decision instead of actual OFDevices. The experimental results show that during migration, the DSSDN stabilizes the load hikes, improves QoS, and increase the end users' utility without much disruptions in the network state.