Hi all,

In iot-benchmark experiment[1], we found that LZ4 outperformed SNAPPY in
terms of (de)compression time and compression ratio in most scenarios. In
the VLDB2022 paper[2], we also found that LZ4 performs no worse than SNAPPY
on real-world datasets.

Therefore, we consider changing the default compressor of IoTDB from SNAPPY
to LZ4. How do you think about it?

option 1: Change LZ4 to the default compressor
option 2: Keep SNAAPY as the default compressor

[1] https://apache-iotdb.feishu.cn/docx/ThmHdPKw9oJv95xvLhWco2WnnLh
[2] https://sxsong.github.io/doc/22vldb-encoding.pdf

-- 
Yanze Chen
School of Software, Tsinghua University

Reply via email to