随着以OpenFlow为代表的多匹配域包分类规则的出现,匹配域数量的不断增加、流表宽度的不断增大以及流表规模的不断膨胀,大大增加了硬件存储的压力.为提高现有三态内容可寻此存储器(TCAM)资源利用率,该文提出一种基于规则集特征分析的匹配域裁剪模型Field Trimmer.一方面基于对规则集中匹配域的逻辑关系分析,实现匹配域的合并,从而减少匹配域的数量;另一方面基于对规则集统计规律的分析,实现匹配域的裁剪,使用部分匹配域来达到整体的匹配效果.实验结果表明,相比于其他方案,该方案在较小的时间复杂度下,能够进一步节省OpenFlow流表的TCAM存储空间需求50%左右;对于常见的包分类规则集,该方案所需的储存空间能够节省40%以上.%With the emergence of multi-field packet classification such as OpenFlow, the increasing number of match fields, continuous growth in bit-width of entries and ever growing scale of rule set all bring much pressure on the storage space in hardware. To improve the utilization of the existing Ternary Content Addressable Memory (TCAM) resources, a match field reduction scheme Field Trimmer is proposed based on the analysis of rule feature. On the one hand, with the analysis of logical relationships among different match fields, some fields can be merged to reduce the number of match fields. On the other hand, with the analysis of statistical features in a rule set, some of the match fields are picked up to achieve the classification function of the whole set. Experiment result shows that with less algorithm complexity, the proposed scheme can save around 50% storage space in the rule set of OpenFlow compared to the best prior art, and about 40% storage space in the popular 5-tuple packet classification rule set.
展开▼
