Hi Chris,
I had the possibility to work with Redshift for a while and I could compare it with Exasol. It was a mixed bag.
From an SQL prospective working with Redshift is like working with a 15 years old PostgreSQL:
- There are even queries that cannot run (and you will face them sooner or later).
- They introduced the possibility to write procedure only recently.
- Permissions and user management within Redshift is very basic (e.g. groups cannot be assigned to other groups). Performance wise we didn't have problems, but the amount of data and concurrent queries was limited.
- Tables needs to be partitioned and sorted (a bit like define an index on the table), you need to manually define if a table is partitioned or duplicated in all nodes.
- There are some system tables inherited from PostgreSQL/Paraccel which contains partial information, the real information are in another table (PG_TABLES_DEF vs SVV_TABLE_INFO).
If you are a DBA you would love Redshift, spinning up a cluster takes very little time (and you can script it with CloudFormation). Backups are done automatically and can be restored quickly, you can also pick specific tables from a backup. Scaling is also quite easy and is managed by AWS. There is a monitoring dashboard to see what is happening and even a Query Advisor. Compared to Exasol Redshift here has the upper hand.
Not sure if you notice but, for me, the interesting features are the ones related to cloud functionalities where they can charge you more (backups or scaling) or keep you in the AWS ecosystem. In fact it is no coincidence that the user management in Redshift is bad, you will be much better off if you use AWS IAM Role instead of groups or the introduction of Redshift Spectrum (Impala), a convenient way to run queries on top of big data formats.
Final words. I would not recommend people to use Redshift if they are not already heavily involved with the AWS ecosystem or have real use cases for a growing and shrinking database (even in that case a product like Snowflake is much more developer friendly than Redshift offering the same pros for the DBAs - there the problems are the costs). But, because Redshift is an AWS product, it is very easy for a medium/small company to start the first cluster and start playing with it (then they will hardly move from there).
