Simple Storage Service

S3 is an Object-Based Storage. Store data as objects rather then in file systems or data block.

Basics

All AWS accounts share the same namespace and each S3 bucket must be globally unique (https://<bucket-name>.s3.<region>.amazonaws.com/<key-name>).

S3 is a key-value store, and it stores a key, value, version ID and metadata (e.g. content-type, last-modified, team-name, etc..).

S3 is an highly available (99.95% - 99.99% depending on the S3 tier) and highly durable (11 9's durability) service.

By default every bucket is private (no public access by default). So by default only the owner can read, delete and update files into a bucket.

You can enable server side encryption on the buckets.

You can define Bucket Policies to define which actions a user can take on the buckets.

You can protect the access using Access Control Lists (ACLs) to define which AWS account can access each resource.

Encryption in Transit: it can use encryption in transit with SSL/TSL or HTTPS
Encryption at Rest (server side encryption SSE)
- SSE-S3: enabled by default, the keys are provided and managed by AWS
- SSE-KMS: the keys are provided by AWS and managed by you
- SSE-C: the keys are provided and managed by you
Client Side Encryption is when you encrypt the file by yourself before uploading it
CORS resource sharing can be allowed to enable a bucket to access resources that are allocated to another S3 bucket

Is a highly available and highly durable storage. Designed for frequent access and suitable for most workloads.

It's stored in at least 3 different Availability Zone.

Designed for infrequently access data, so data that you may access a few times a month, but not on a daily bases.

Great for long term storage, backups and disaster recovery files.

It provides Rapid Access, you pay to access the data (low per-GB storage price and a per-GB retrieval fee).

It's stored in at least 3 different Availability Zone.

Same as the S3-IA but it's available only in one AZ, but costs 20% less then a regular S3-IA.

There are 2 Glacier Options: Glacier and Glacier Deep Archive.

They are both very cheap and designed for data that needs to be accessed once per year, so good for archiving data.

To access data from the normal one it can take from 1 minute to 12 hours, while for the second has a default retrieval time of 12 hours.

Automatically moves you data on the most cost-effective tier based on how often you access the data.