Thanks this was a great video! I'm pretty new to system design and was trying to wrap my head around the difference between column-based and row-based databases and the visual and examples in this video were perfect :)
Great Explanation !! That visual representation of storing data in clickhouse gives clear picture about its behind the scene logic earlier I used to struggle visualizing how it handles data despite knowing its column way to store data but now its crystal clear Thanks :)
What about pricing? Isn't postgress free self hosted while clickhouse not? Self hosted vs not isn't click house saas so you need to send them all your data and if you have also local self hosted data that means you would now have two sources of data. While postgress you can self host it.
No. clickhouse free open-source too. saas clickhouse-cloud just new standalone product if you need one. But new self hosted versions is still free and open-source.
This sounds using an ELT (like with Dbt) approach must be horrible with clickhouse. Since you load everything in and the mutate the data. You said Mutation take very long here.
4:58 i'm pretty sure that this isn't really good example, if you want to built clickhouse-alike-materialized-view on postgresql you have to use stored procedures instead of materialized view, it's pretty much wrong feature naming on clickhouse side. And i don't even like postresql. 6:30 clickhouse sharding is pretty much equal to manual transmission(as an example: you have to built whole cluster before inserting any data because there's no resharding mechanism at all, you can only recreate tables and reincert all of the data, it's cool that this mechanism exists, but it's far from being perfect) in comparison to databases which actually handle it by themselves. True "zero hassle" sharding is only available in proprietary "clickhouse cloud" build which you can't have on-premise unfortunately.
Thanks this was a great video! I'm pretty new to system design and was trying to wrap my head around the difference between column-based and row-based databases and the visual and examples in this video were perfect :)
Glad it was helpful.
Great Explanation !!
That visual representation of storing data in clickhouse gives clear picture about its behind the scene logic
earlier I used to struggle visualizing how it handles data despite knowing its column way to store data
but now its crystal clear
Thanks :)
Excellent video.
What about pricing? Isn't postgress free self hosted while clickhouse not? Self hosted vs not isn't click house saas so you need to send them all your data and if you have also local self hosted data that means you would now have two sources of data. While postgress you can self host it.
No. clickhouse free open-source too. saas clickhouse-cloud just new standalone product if you need one. But new self hosted versions is still free and open-source.
This sounds using an ELT (like with Dbt) approach must be horrible with clickhouse. Since you load everything in and the mutate the data. You said Mutation take very long here.
Clickhouse is kind of fast persistent cash. Never use it for transformations, only as fast access to data
4:58 i'm pretty sure that this isn't really good example, if you want to built clickhouse-alike-materialized-view on postgresql you have to use stored procedures instead of materialized view, it's pretty much wrong feature naming on clickhouse side. And i don't even like postresql.
6:30 clickhouse sharding is pretty much equal to manual transmission(as an example: you have to built whole cluster before inserting any data because there's no resharding mechanism at all, you can only recreate tables and reincert all of the data, it's cool that this mechanism exists, but it's far from being perfect) in comparison to databases which actually handle it by themselves. True "zero hassle" sharding is only available in proprietary "clickhouse cloud" build which you can't have on-premise unfortunately.
That's called replication then, not sharding
Great Video. Thank you !
Great explanation :)
woo!