Smart Flow Table Compression

Summary

Multi-field packet classification is a crucial component in modern software-defined data center networks. To achieve high throughput and low latency, state-of-the-art algorithms strive to fit the rule lookup data structures into on-die caches; however, they do not scale well with the number of rules. We present a novel approach, which uses nerual network instead of traditional data structure.

• Introduces the prior knowledge of “mask length” to the neural network, does not require its own learning, and does not generate other additional overhead
• When the rules are updated, only the neural network corresponding to the tuple needs to be retrained. Since the size of a single neural network is small, it can support fast incremental update
• The process of package inference can be parallelized, which speeds up package inference