دانلود رایگان مقاله تایید ناهماهنگی حالت SDN در اوپن استک

Abstract

In Software Defined Networks (SDN), users manage network services by abstracting high level service policies from lower level network functions. Edge-based SDN, which relies on end hosts to implement lower level network functions, has been rapidly developed and widely adopted in cloud system. A critical challenge in such an environment is to ensure that lower level network configurations, which are distributed in many end hosts, are in sync with the high level network service definitions, which are maintained in the central controller, as state inconsistency often arises in practice due to unreliable state dissemination, human errors, or software bugs. In this paper, we propose an approach to extract, analyze the network states of OpenStack from both controller and end hosts, and identify the inconsistencies between them across multiple network layers systematically. Through extensive experiments, we demonstrate that our system can correctly identify network state inconsistencies with little system and network overhead, therefore can be adopted in large-scale production cloud to ensure healthy operations of its network services.

11. Conclusion

In this paper, we studied the problem of network state verifi- cation in Edge-based SDN. We take OpenStack as our study focus. Our solution consists of data extraction, state parsing and state verification at L2, L3 and L4 layers. To reduce the parsing time for L4 state, we proposed a two-level comparison design and developed a hierarchical BDD caching algorithm. We further studied and proposed continuous verification schemes to handle the frequent SDN state changes. Through extensive experiments on our local testbed, we demonstrated that our verification system can effectively detect a variety of configuration inconsistencies in a dynamic cloud setting, with low computation and memory overheads. Although we focus on studying OpenStack, our solution framework and the techniques within this paper could be easily adopted in solving similar problems in other Edge-based SDN systems as well.