Redshift is a data warehouse from Amazon. Its a fork of PostgreSQL that writes data in a column oriented format, making it much faster at analytical queries.
Redshift is very popular due to its availability within AWS and the fact that it works quite well for most use cases.
As far as I understand it, there are two main versions of Redshift:
The original version of Redshift is provisioned. This means you must select compute and storage resources when starting an instance of the service and pay for them regardless of usage.
A newer version of Redshift is serverless. This follows the recent trend of separating storage and compute resources, like Snowflake or Bigquery.
Both of these versions present the same (mostly) PostgreSQL compatible API, but use very different storage engines. If you can stomach the up-front flat fee of $2,160 per month, I recommend using the serverless version of Redshift, as it is a much more modern design.
In my opinion, serverless Redshift is a good option for large organizations that are already on AWS and want to minimize data integration pain. Everyone else should lean towards Snowflake or Bigquery.
Pros
Fantastic integration with other AWS services, such as their Zero ETL feature.
Very good performance for analytical queries.
Compatible with PostgreSQL clients.
Cons
The provisioned version doesn’t separate storage from compute. This is not a modern design pattern, and in my opinion no one should adopt this kind of data warehouse anymore.
The Redshift UI is meh.
The “serverless” version costs a flat $3 an hour in addition to any compute and storage resources you use. In my opinion, you shouldn’t call something serverless when it costs $2,160 a month before you even start using it.
I will make one caveat: If you are a big organization that doesn’t care about $2,160 per month, their pricing is fine after that initial cost.