S3 Standard vs Intelligent Tiering – What’s the difference

What’s the difference between S3 Standard vs Intelligent Tier? Learn the difference in performance characteristics and cost in this blog post.

What is Standard Tier?

Standard Tier is the default, general purpose tier of object storage intended for a variety of use cases. It’s been around since the launch of Amaazon S3 in 2006 and hasn’t changed much.

The main features offered are:

1. High Availability (99.99% SLA)
2. Low Latency and High Throughput – even extremely high throughput applications.
3. High durability aka data loss protection (99.99999999999% durability guarantee)
4. Protected from availability zone outages since data is copied across multiple AZs

Overall, Standard Tier is a general classification that is a good ‘all arounder’.

What is Intelligent Tier?

Intelligent tier was announced in 2018 and allows users to save on S3 costs automatically when access patterns change to your data. Performance is not compromised and there is no additional work on the user’s part beyond initial configuration.

Its important to note that by using intelligent tier, you need to pay for a small object monitoring fee to keep track of access patterns to your data; this is maintained and automated by AWS itself. The actual cost depends on the amount of objects you’re monitoring – more objects costing more and less costing less, obviously.

Intelligent tier shuffles objects between two tiers based on high or low access to data. Data first uploaded into the intelligent tier is assigned to the “high access tier”, since data most recently uploaded is more likely than not to need high performance and frequent access. After a period of 30 days of infrequent access, data is moved to the low access tier. This is where the cost savings come into play. If for whatever reason access to that object picks back up, AWS will move it back into the high access tier.

Its important to note that objects under intelligent tier (regardless of which high or low tier they are currently assigned to) have very similar if not the same performance / latency characteristics. AWS doesn’t exactly specify the difference, just that they are pretty much the same.

There’s also an option of using the Archive and Deep Archive access tiers (this is basically the same as Glacier / Deep Glacier). I wouldn’t suggest this though for those of you looking for low latency access to your data.

All in all, intelligent tier has the following characteristics:

1. Automatic cost optimization by shuffling data between two tiers
2. Same low latency and high performance as Standard
3. Slightly lower availability than standard tier (99.99% for Standard, 99.9% for intelligent)
4. Small monthly monitoring / auto tiering cost.
5. Option to further extend cost savings by promoting to Archive tier

Standard vs Intelligent Cost Comparison

Lets put together a quick hypothetical example to demonstrate the cost difference between these two beasts. Below are the conditions for the comparison:

Conditions

1. 100GB of Data Storage
2. 1000 GET / PUT Requests per month
3. For intelligent tier, 50% of data in frequent tier/high tier, 50% of data in infrequent tier
4. For intelligent tier, 5000 objects being monitored
5. 0 data transfer out of AWS

With these conditions, the grand total when using Standard Tier comes to $2.31 USD per month.

For Intelligent Tier, this comes out to $1.79 USD per month.

That’s a cost savings of nearly 25% by going with intelligent over standard.

To run your own cost experiments, you can use the S3 pricing calculator here.

Summary

Standard and Intelligent tiers are very similar storage classifications of your S3 objects. Using Intelligent, you can take advantage of S3’s shuffling of your data into different tiers based on your access patterns.

I highly suggest those of you with unknown access patterns to give Intelligent Tier a go. It can greatly reduce your costs of using S3 with almost no downsides.