richraposa

5 ms·

1.
by richraposa 3y ago
...says the completely anonymous internet guy. I'm laughing as much as you are. I wasn't trying to imply anything negative about DuckDB with my post - was just sharing how ClickHouse does the same thing. FWIW: the blog author adde
2.
by richraposa 3y ago
ClickHouse just reads the values from the one column being summed: https://www.markhneedham.com/blog/2023/11/15/clickhouse-summ...
3.
by richraposa 3y ago
Nice. I did the same thing on ClickHouse - took 11 seconds: SELECT sum(size) FROM url('https://huggingface.co/datasets/vivym/midjourney-messages/resolve/main/data/0000{01..55}.parq
4.
by richraposa 3y ago
CGW = Can't Go Wrong...love it
5.
by richraposa 3y ago
ClickHouse 10 years before DuckDB existed: SELECT * FROM url(' https://example.com/*.csv ')
6.
by richraposa 3y ago
I suppose if you had data in a format that DuckDb doesn't work with, like Protobuf, Avro, ORC, Arrow, etc. ClickHouse reads and writes data in over 70 formats
7.
by richraposa 3y ago
ingestion of these events during peaks at 500k events per minute. You can't ingest them individually into Clickhouse or most other databases. Turn on async_insert or use a Buffer table engine and you can easily insert them individual
8.
by richraposa 3y ago
Cloudflare famously uses ClickHouse for web analytics - inserting over 6M rows per second: https://blog.cloudflare.com/http-analytics-for-6m-requests-p...
9.
by richraposa 4y ago
It's definitely cool to be able to query data in place instead of inserting it into a table. You can use clickhouse-local to do the same thing with JSON files (and with dozens of other data formats): https://clickhouse.com&#
10.
by richraposa 4y ago
You should try it on ClickHouse Cloud - the pricing is not based on bytes, so the resulting cost would be near zero compared to $1M on BigQuery. And it would probably execute faster... :) (Disclaimer: I work at ClickHouse)