One of the biggest reasons we moved from MySQL to Snowflake was to improve query performance. Selects that would basically never come back on our poor, battered MySQL server could be completed in minutes on Snowflake. What we didn’t realize at the time was we’d be able to also eliminate a fair amount of hadoop mapreduce jobs – we could get the same results through views and save a ton of time!
It’s pretty cheap.
Snowflake decouples your compute and storage, so we’re able to scale up / down depending on demand. This is basically the dream of cloud computing come true.
It’s easy to use.
Snowflake queries are in SQL! Not having to learn a new language and syntax has been huge, especially for our team members that are not engineers. Snowflake has comprehensive documentation online as well for all of their custom stuff. You can use their web interface for queries, command line tool, or even through SQL Work Bench. Integrating with python was a breeze through the JDBC driver.
Wait, there’s more!
There are a lot of other great features of Snowflake that probably deserve their own blog posts, but to name a few… native timestamp support, custom “schemas” (create fast lightweight copies of your data for safe testing with failovers to select from default), time-travel (query what your data looked like at a certain time) and semi-structured data (load JSON or AVRO directly from S3).
If you want to read more about Snowflake, check out their site! Or just wait for us to write some more blog posts as we continue to use and love it. 🙂