Amazon RDS

It is a pay-per-use system where you are only charged for the number of transactions that occur. Still, there is also a subscription system called Reserved Instances (RI), so instances that generate transactions frequently use this system to reduce costs significantly. There are three types to choose from: one-year prepayment, one-year prepayment, monthly payment, and no prepayment, depending on the kind of discount. Pay-per-use is basically cheaper for cases where transactions do not occur frequently, depending on the situation, so you can optimize the charge by using different payment methods depending on the application.

I would like you to provide a concrete fault test solution. If you do not have a high level of knowledge about RDS, you cannot be relieved that you cannot perform useful failure tests assuming a failure occurs.

AWS RDS allows us to deploy databases easily and efficiently with great resilience, scalability, and security - but without having to deal with managing a server. RDS makes it easy to spin up a new database, have multi-availability zone replicas, back it up, and do many other things. It's also dead easy to upgrade software versions - simply let AWS look after it.

Being a managed database service there are restrictions; for instance, with SQL Server you don't have the sysadmin role and you can't set any trace flag you want. You can adjust a number of parameters in the RDS console but only those which AWS have explicitly added support for. There are other caveats and restrictions for other DBMS products too.

Another thing I don't like is when you change your instance size it takes a long time - but at least there's no downtime. Also, tooling is restricted.

A lot of great tools to dig into performance and query tracing won't work with the AWS caveats so you have to use the AWS RDS console and logs and it can be more tedious to really drill into application performance bottlenecks.

Requires minimum knowledge or time to set up the database.
Everything you need is basically there.
Documentation is fantastic.
It just works and takes all the burden of infrastructure management from you.
A handful of engines, version, and extensions to them are supprted.
Integrated seamless to AWS infrastructure.

Price. It will literally eat 7/8 of your infrastructure budget.
Metrics lies. Performance insight simply does not work. Note, this is an opinion of somebody who paid for AWS Enterprise support (sigh, that's me)
Backups are not backups per se, this is filesystem rsync. So it is easy to end up with broken DB.
Support is very poor.

Aurora is a great port of mySQL - very compatible and super fast
Prices are always going one direction: down
Scalable way huge with just a few clicks
Automated backups, patching, upgrades
You can still do a lot of customization using PL/MYSQL
The instances just stay up and running - becomes one less thing to worry about
You have options: both mySQL and Postgres

It's not amazingly easy to update the various variables that enable you to configure your mySQL instance.
I wish it was a bit easier to get monitoring that would give you more granular insight into what's causing issues.
You don't have quite as much flexibility and control over special packages you might install to do special stuff (calculating the Levenschtein distance between words, for example).

I like how easy it is to scale RDS up and down alongside its monitoring features which can be conveniently accessed via AWS console.

The burst balance limit for the underlying EBS disk gave us lots of problems. I wish this feature was more user-friendly.