wanghenshui / wanghenshui.github.io

my blog, please do not fork
https://wanghenshui.github.io
Other
4 stars 1 forks source link

blockio #144

Open wanghenshui opened 1 week ago

wanghenshui commented 1 week ago

https://smalldatum.blogspot.com/2024/07/searching-for-regressions-in-rocksdb.html

wanghenshui commented 1 week ago

github.com/facebook/rocksdb/pull/12937/files

wanghenshui commented 1 week ago

https://ronekins.com/2024/01/16/how-to-reduce-linux-block-storage-io-sizes/

max_hw_sectors_kb (read-only) > max_sectors_kb (read/write)

注意max_sectors_kb大小

一个sector 512B 4MB (8192512B) avgrq-sz sector-sz

https://kernel.dk/when-2mb-turns-into-512k.pdf

IO调度,是存在组合的

bio 256 个page 4k

多个bio 组合成batch

wanghenshui commented 1 week ago

https://helloyoubeautifulthing.net/blog/2021/01/03/parquet-format/

https://tech.ipalfish.com/blog/2020/12/08/apache_arrow_summary/

https://smalldatum.blogspot.com/2024/07/searching-for-regressions-in-rocksdb.html

wanghenshui commented 1 week ago

https://plantegg.github.io/2019/06/21/%E5%B0%B1%E6%98%AF%E8%A6%81%E4%BD%A0%E6%87%82TCP--%E6%80%A7%E8%83%BD%E4%BC%98%E5%8C%96%E5%A4%A7%E5%85%A8/

wanghenshui commented 6 days ago

https://fee-mendes.gitbook.io/scylladb-mc-compare#tests-and-results

seastar简单场景是没优势的。

停止对技术名字的神圣化

wanghenshui commented 6 days ago

https://www.scylladb.com/2017/07/31/database-caches-not-good/

多一跳 成本 可用性 应用场景复杂化

外部缓存破坏后段数据缓存。这个点可以,需要重放 热key 点查,防止冷冲击

外部缓存不安全 存疑

外部缓存没有更好的和数据库后段数据联动,这个和前面的 有点冲突

防止冷启动/proxy预热是比较重要的点

wanghenshui commented 5 days ago

https://nan01ab.github.io/2019/02/WAL-Optimizations.html

wanghenshui commented 5 days ago

https://datafusion.apache.org/blog/2024/09/13/string-view-german-style-strings-part-1/

https://datafusion.apache.org/blog/2024/09/13/string-view-german-style-strings-part-2/

https://www.tunglevo.com/note/an-optimization-thats-impossible-in-rust/

wanghenshui commented 5 days ago

https://www.datadoghq.com/blog/engineering/llms-for-postmortems/

https://www.datadoghq.com/blog/engineering/timeseries-indexing-at-scale/

wanghenshui commented 5 days ago

https://blog.cloudflare.com/sqlite-in-durable-objects/

wanghenshui commented 5 days ago

https://cedardb.com/blog/colibri/

wanghenshui commented 5 days ago

https://medium.com/@adamprout/categorizing-how-distributed-databases-utilize-consensus-algorithms-492c8ff9e916

wanghenshui commented 2 days ago

https://news.ycombinator.com/item?id=40972099

http://web.stanford.edu/class/ee380/Abstracts/200205-slides.pdf