Abstract

Software-defined networking (SDN) enables flexible fine-grained networking policies by allowing the SDN controller to install packet handling rules on distributed switches. The behaviour of SDN depends on the set of forwarding entries installed at the switch flow table. The increasing number of traffics from the proliferation of the Internet of Thing (IoT) devices increase the processing load on the controller and generates an additional number of entries stored in the flow table. However, the switch flow table memory (TCAM) cannot accommodate many entries. Packets from multimedia flows are usually large in size and thus suffer processing delay and require more flow set up requests. The SDN controller may be overloaded and face some scalability problems because it supports a limited number of requests from switches. OpenFlow uses timeout configuration to manage flow setup request. The conventional fixed timeout cannot cope up with the dynamic nature of traffic flows. This paper controls the frequent flow setup requests by proposing an adaptive and hybrid idle–hard timeout allocation (AH-IHTA). The algorithm considers traffic patterns, flow table usage ratio, and returns appropriate the timeout to different flows. The performance evaluations conducted have shown a 28% and 39% reduction in the flow setup request and flow eviction, respectively.

Highlights

  • Software-defined networking (SDN) has recently gained popularity among researchers, industry, and even carrier-grade network due to simple and programmable network management

  • This paper proposes an adaptive-hybrid idle and hard timeout allocation (AH-IHTA) scheme to improve the efficiency of the flow table and reduce the controller overhead without modifying the SDN

  • Due to the poor nature of a flow table update operation, a certain amount of free space needs to be preserved (20%) to speed up the process, and strictly control the increment in entry lifetime. This way, the AH-IHTA module set the Tdel flag to notify data flow eviction module (DFEM) to explicitly remove a less important entry considering the traffic characteristics: AH-IHTA is presented in Algorithm 1

Read more

Summary

Introduction

Software-defined networking (SDN) has recently gained popularity among researchers, industry, and even carrier-grade network due to simple and programmable network management. Devices had significantly increased the amount of traffic flows in the internet, and this causes the traffic pattern to exhibit different variability in terms of arrival and duration [6] This may surge the processing load on the controller because of the need to process every incoming flow and install the corresponding flow entries in the switch flow table. These concerns increase the level of packet-in generation due to inappropriate timeout and an ineffective flow entry eviction policy. This paper complements the previous work by extending the previous experiment to consider various data flow classification and more eviction mechanisms were added to show the effectiveness of the proposed evictions mechanism

Timeout Mechanism
Eviction Mechanism
Related Work
Traffic Flow Pattern
4: Modules of the proposed solution
Output
Experimental Result and Discussion
Result for Idle and Hard Timeout
Result for Idle and Hard Timeout with Different Flow Table Size
Result for Data Flow Entry Eviction
Findings
Conclusions
Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call