By Daniel do Nascimento

Since the 1990s, the cost of storage per MB has been decreasing very rapidly, and this assumption that we don't have to think about the cost of storage has permeated most developers minds for a while.
But in the age of big data and cloud services, this issue has come back well, especially when we use pay-as-you-go services like S3, the next month's bill may remind us of a large amount of data that we forget to clean, or it is not being used.
Fortunately for us, AWS has some tools to deal with this situation and save some money.

Options

S3 Standard : Designed for frequently accessed data.

S3 Intelligent-Tiering : Data with changing or unknown access patterns.

S3 Standard-IA : Designed for long-lived, infrequently accessed data.

S3 One Zone-IA : Designed for long-lived, infrequently accessed, non-critical data.

S3 Reduced redundancy : Frequently accessed, non-critical data.

S3 Glacier Instant Retrieval : Designed for archive data accessed once a quarter with instant retrieval in milliseconds.

S3 Glacier Flexible Retrieval : Designed for infrequently accessed, archived critical data.

S3 Glacier Deep Archive : Long-term data archiving with retrieval times within 12 hours.

As you can see there are many options, the problem is that most companies using S3 don't do all the research on the correct tier to choose when they upload the objects, and end up using S3 Standard for everything, primary because storage should be simple (Simple Storage Service), not (Complicated Storage Service).

The problem with this approach is obvious, we tend to pay for a service we don't really need, but aws offers between these tiers an option that solves most cases at a good discount, and this should be the new standard in my opinion.

The S3 Intelligent-Tiering moves objects from S3 Standard to S3 Infrequent Access, for objects that are not used in 30 days, and moves the objects not used for 90 days to Archive Instant Access Tier(Same characteristics as S3 Glacier Instant Retrieval), saving up to 68% on these objects.

How it works?

S3 Intelligent-Tiering monitors the age of your objects (larger than 128KB), and since they achieve a determined life(in days) without access, the Intelligent-Tiering change the object tier configuration, putting the object in appropriate tier e also retrieving from the tier when the object is accessed, so continuously optimizing your object tier configuration.

How much does it cost?

In order to allocate the objects in the correct tier, S3 Intelligent-Tiering keeps track of the age of your objects and this creates the additional cost. But this cost is very low when compared to the savings that he generates.
The additional cost of S3 Intelligent-Tiering Monitoring and Automation today(february/2023) is 0.0025$ per 1000 objects. You can check out the total cost using: S3 Pricing

How to use S3 Intelligent-Tiering

There are 2 ways to configure your objects to use Intelligent-Tiering, the first and simpler way is just specify the INTELLIGENT_TIERING in the PUT API request header. Ex:

But, as you guessed, this method only works for one object at the time. What if you want to convert all the objects that already are in a bucket, to S3 Intelligent-Tiering? Then you have to create a S3 Lifecycle Rule in your bucket, you can apply this rule to the bucket or folder/prefix level.

You can go further with your optimization and activate the Archive Access Tier and Deep Archive Access Tier, but these two options require you to make a better analysis on your objects and use, because they don’t allow for real-time access, they have retrieve times from minutes to hours.
This simple configuration doesn't substitute a careful analysis of your application needs, but helps you to save big without much complexity in configuration.