1. Home
  2. Relational Databases
  3. Amazon RDS Reviews
  4. User Review of Amazon RDS
Great for a quick analytics database, wouldn't use it for production though
March 28, 2017

Great for a quick analytics database, wouldn't use it for production though

Valeri Karpov | TrustRadius Reviewer
Score 10 out of 10
Vetted Review
Verified User

Overall Satisfaction with Amazon Relational Database Service

Amazon Relational Database Service (RDS) hosts a Postgres instance that we send our production MongoDB data into. We then use metabase (http://www.metabase.com/) on top of RDS for analytics. MongoDB hosts all our production data and is the source of truth, but we copy that data into RDS because most of our non-technical staff are not proficient with MongoDB's aggregation framework and because we don't want to risk an errant query hurting our production database's performance.
  • Quick and easy hosting for SQL databases. Setting up Postgres locally is a nightmare and we don't have experience managing it, RDS has made it easy.
  • Easy to set up tools like MySQL to pipe your real production data into RDS.
  • Gives you a nice neat Postgres URI that you can point metabase, wagon, etc., at for analytics.
  • RDS is expensive compared to running a self-hosted SQL database on EC2 (once you get past the generous free tier). Currently the cheapest on-demand RDS instance costs $0.041 / hour, whereas EC2 t1.nano instances cost $0.006 / hour. $29/mo seems expensive given that we have no need for blazing performance or multi-availability-zone durability.
  • I don't believe traditional RDBMS have any business being production databases in this day and age, so I find RDS' focus on performance and durability to be wasteful and distracting.
  • Like most AWS products, the documentation is exceptionally complex and difficult to navigate, and the UI displays an awful lot of superfluous information.
  • Combined with metabase, RDS has enabled numerous non-technical members of our team to run and share sophisticated queries.
  • The engineering time spent on worrying about RDS has been minimal. We set it up once over a year ago and haven't touched it since.
  • We're relying on metabase and RDS more for operations-related dashboards, because it's generally faster to iterate on a simple query than a complex web application.
Redshift is massively scalable but has some limitations that we weren't willing to accept (no JSONB). It also has its own distinct flavor of SQL, and there isn't as much content online about Redshift's flavor of SQL versus postgres'. In the end, we just didn't need to kind of scale that redshift was designed for, so we went for RDS, because it was cheaper and simpler.
I think RDS is a great solution for hosting an SQL analytics database if you're not big enough to justify using Redshift and want to avoid Redshift's limitations. Getting set up is pretty easy once you've digested the somewhat unclear documentation, and I've had no complaints about the service's reliability. Setting up and managing Postgres is a nightmare, so I was thrilled that Amazon could do it for me. However, I wouldn't trust it to run your production data, because I wouldn't recommend a relational database for running your production in general.