The Research On Efficient Storage And Accelerated Lookup For OpenFlow Large-Scale Flow Table

Software Defined Networking is widely considered as one of the most promising direction toward future network,and currently OpenFlow is the most prevalent southbound interface protocol in SDN deployments.When SDN is deployed in large-scale networks,such as wide area network,data centers,OpenFlow switches are faced with heavy performance bottlenecks of packet forwarding due to the expanding scale of OpenFlow tables and the increasing expenses of flow table lookups.This paper is thus motivated to propose a differentiated storage and accelerated lookup architecture for large-scale flow tables(DAFT),which is used to ease the storage pressure of OpenFlow large-scale flow table and improve the flow table lookup performance.The main research work of this paper is as follows:(1)To address the problem of OpenFlow large-scale flow table storage,we first investigate into the impact of wildcards in match fields on the packet-in-batch feature within a flow based on network traffic locality.Then,packet flows are dynamically distinguished into active ones and idle ones in terms of their short-term states.Subsequently,we store the match fields of active flows and idle flows respectively in TCAM and SRAM,and the content fields of both types of flows in DRAM,to effectively relieve the insufficiency of TCAM capacity.Finally,we evaluate the performance of our proposed flow table storage architecture with real network traffic traces by experiments.The experimental results indicate that our proposed storage architecture with the active/idle flow differentiation obviously outperforms the traditional one applying the elephant/mice flow differentiation in terms of TCAM hit rates.(2)To accelerate the lookups of SRAM flow table,we mitigate the performance bottleneck of SRAM flow table lookups in two ways.On the one hand,we self-adaptively adjust mask order by applying the Move-Ahead-1 heuristic,to reduce the number of failed mask probing of subsequent packets,in virtue of the non-uniformity property of mask hitting.On the other hand,we predict tuple search failures by employing counting bloom filters(CBF),to bypass the traversal on the corresponding sub-flow tables,in terms of the property that most mask probing tends to fail.Finally,we evaluate the lookup performance of our proposed accelerated lookup scheme for large-scale OpenFlow table by experiments with backbone network traffic traces.The experimental results indicate that our proposed scheme obviously performs better than the traditional,in terms of average search length in SRAM and average access time on flow tables,which effectively promotes packet forwarding performance of OpenFlow switches.(3)To reduce lookups energy consumption of TCAM flow table,we first proposes an energy-saving OpenFlow flow table lookups architecture,and uses low-power RAM to cache acitve flows of TCAM,making most packets directly hit cache instead of TCAM.Then,we designes a named FlowSparsegrid flow cache,whose number of cached flows can be flexibly adjusted by altering threshold of active flow identification parameters,and futher achieves the best balance between flow table lookups performance and energy cost.In addition,the cross-chain structure of FlowSparsegrid can effectively store or adjust the mapping relationship between the cached flows and TCAM flows,ensuring all packets can always hit the highest priority flow table entry,and effectively solving the inconsistency problem of flow table lookups results,caused by limits of rule dependency.