Data Center Application Traffic Distribution via Programmable Switches

The patent segments the rule set into multiple priority groups (first group with higher priority, second group with lower priority). Packets are matched against rules in the first group first, and only if no match is found are they evaluated against the second group. This segmentation reduces the average number of rules each packet must be matched against, thereby reducing latency and jitter while preserving comprehensive traffic routing capability.

The patent divides the rule processing workload into segmented priority groups. By organizing rules into first and second priority groups, the system ensures that packets matching high-priority rules are processed quickly without requiring evaluation of all rules. This segmentation maintains rule processing speed even as the total number of rules increases to handle greater workload capacity.

Instead of migrating virtual machines to address traffic distribution issues (which consumes significant resources), the patent inverts the approach by migrating rules to different priority groups based on traffic patterns. This rule-based approach achieves similar resource allocation efficiency without the resource-intensive VM migration process, thereby improving productivity while reducing device complexity.

The patent segments rules into priority groups where the first group contains high-priority rules and the second group contains lower-priority rules. This segmentation ensures that high-traffic workloads are quickly routed via high-priority rules without being blocked by numerous low-priority rules, maintaining both service quality for low-traffic workloads and lookup speed for high-traffic workloads.