Approximating Fair Queueing on Reconfigurable Switches
summary
Implementation of approximate Fair Queueing on reconfigurable switches uses per-packet processing and mutable on-switch state to emulate bit-by-bit round robin with a count-min sketch variant for per-flow rounds, a bytes-per-round quantum, and a Rotating Strict Priority dequeue implemented via a small set of FIFO queues. Prototype on a Cavium NPU and simulations on a leaf-spine topology show using 12–16 FIFOs and <10% extra resources yields 2–4x faster average short-flow completion and 4–8x better 99th-tail latency versus TCP/DCTCP.
AI-written summary · about
chapters (14)
- 0:00Intro
- 0:37Congestion Control Today
- 1:32Fair Queueing : in-network enforcement
- 2:46Fair Queueing without per-flow queues
- 6:53Realizing Fair Queueing on Reconfigurable Switches
- 7:59Our approach: Approximate Fair Queueing
- 8:41Storing Approximate Flow Counters
- 11:28Buffering Packets in Approximate Sorted Order
- 13:09Rotating Strict Priority (RSP)
- 13:54Realizing an RSP Scheduler
- 15:04Summary of Techniques
- 17:09Evaluation
- 17:21Prototype Implementation
- 19:08Simulation Results: Comparison to Fair Queueing
related talks
12mTwenty Years After: Hierarchical Core-Stateless Fair Queueing
16mManaging Congestion Control Heterogeneity on the Internet with Approximate Performance...
22mProgrammable Calendar Queues for High-speed Packet Scheduling
16mGearbox: A Hierarchical Packet Scheduler for Approximate Weighted Fair Queuing
16mPRED: Performance-oriented Random Early Detection for Consistently Stable Performance...
23mSP-PIFO: Approximating Push-In First-Out Behaviors using Strict-Priority Queues
13mSifter: An Inversion-Free and Large-Capacity Programmable Packet Scheduler
16mBackpressure Flow Control
