2333xb commented on issue #13927: URL: https://github.com/apache/apisix/issues/13927#issuecomment-6136122668
Thanks for the clarification. My use case is HTTP response compression at the APISIX gateway for browser clients. My motivation is zstd’s promising balance of compression speed, compression ratio, and decompression speed. The published [[zstd benchmarks](https://github.com/facebook/zstd#benchmarks)](https://github.com/facebook/zstd#benchmarks) show substantially faster compression and decompression than zlib and Brotli at the tested low compression levels, with comparable or better compression ratios. For on-the-fly compression at the gateway, this suggests an opportunity to reduce CPU cost while maintaining good bandwidth savings. This is the main benefit I would like to see APISIX support. I understand that the actual benefit depends on payloads and compression settings, and I do not yet have APISIX-specific benchmarks or fixed performance targets. The intended clients are browsers that advertise `zstd` in `Accept-Encoding`, with existing gzip/Brotli negotiation preserved for other browsers. An opt-in implementation would be a useful starting point, evaluated for lower CPU usage at comparable compressed sizes and its impact on response latency. Would this be sufficient motivation to consider zstd support? -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
