TECC: Towards Efficient QUIC Tunneling via Collaborative Transmission Control
Talkat 21st USENIX Symposium on Networked Systems Design and Implementation (NSDI '24) · 2024-04-16..04-18·
summary
TECC implements a feedback framework where the tunnel server sends RTT, queue length, retransmit rate and pacing to the client so the end-to-end QUIC sender recalculates a safe server sending rate (SRT = trt − Q/θ − dRT·trt/θ with a max reduction cap). TECC synchronizes congestion control between tunnel and inner connection, is evaluated with mahimahi emulation and large-scale A/B tests, reduces flow completion time by ~30% on average and 53% at the 99th percentile in emulation, and cuts RPC latency 3.9% (avg) / 13.3% (99th) with <=2% setup overhead.
AI-written summary · about
chapters (12)
- 0:00TECC and QUIC tunneling intro
- 1:26QUIC tunnel use cases and RPC
- 2:31Stream vs datagram tunnels in MASK
- 3:17R-transmission impact on datagrams
- 4:24Nested congestion control challenges
- 6:31Collaborative transmission control design
- 7:37Server sending rate in TCC
- 9:44Experimental setup and networks
- 10:26Performance under packet loss
- 11:07Delay impact on TCC
- 11:50Overhead and deployment results
- 12:57Summary and future
related talks
20mETC: An Elastic Transmission Control Using End-to-End Available Bandwidth...
19mThe Future With QUIC
16mR-TCP: A Framework to Optimize TCP Performance Over Rate-Limiting Networks
42mMQTT over QUIC: Opportunities and Challenges
16mPowerTCP: Pushing the Performance Limits of Datacenter Networks
14mInternet Connection Splitting: What’s Old is New Again
23mACK to the Future: TCP in 2025
31mTuning Linux TCP for data-center networks