Before I started making changes to the infrastructure the rule was simply go one step up in server size, everytime there was an spike and it was confirmed that the database was the root case the server size was increased and at some point we had a single 8xlarge instance for our main database.

In the past there was no logging system implemented and the actions were based only on resources usage. We started getting better insight after our Datadog implementation in AWS, we set alerts for resources usage and we also enabled logs from the database including the slow queries logs one.

You might be thinking, Who deploys a database in a public subnet? Even worst, it was deployed in the AWS default public subnet with publicly accesssible enabled, this is what happends when services are deployed just because without thinking on the implementation but just business requirements.

At first it was a bit complex to simply change the publicly accessible to disabled as there were a few external data analytics tools that were pulling information from the database every few hours, the first time I tried so, I endup in a call with the CTO and the Data Analytics team asking me why I disabled the public access. Simple answer - Security.

In the past we were using the VPN solution from AWS and even though pero hour seems to be cheap, the truth is the costs scale quickly when the whole team needs to be connected to it and we were an small team.

At some point I was investigating an spike on billing and this is what a I marked as possible improvement moving from AWS VPN solution to OpenVPN:

Cost savings is a pretty important part of working with AWS, knowing your business plan helps you to plan in advance and take advantage of the reserve instances plan from AWS, In a previous post we were discussing the power of reserved instances for compute services and this case is not different, we will see an example on how a commitment can help to reduce your database expenses.

Although this is not the only way to reduce database costs it is a quick and efective way to achieve it, in a few steps you will be able to set it and the next day it will be running. To reserve your instances go to RDS > Reserved Instances > Purchase Reserverd DB Instance. This is an example of how the reserved instances look after purchase:

At some point on 2023 we were discussing about costs and pagespeed, I quickly pointed out that the slow performance and spike in bandwidth cost was caused by the usage of .gif images, at first the team was reluctant about this but a few days later I came up with a report provided by the Amazon S3 service in which you can get save an access log of the hosted files and later process it by using Amazon Athena.