This article is a quick overview of Amazon S3, one of Amazon Web Services.
What is S3?
S3 stands for "Simple Storage Service, a Cloud-based intelligent object storage service.
In essence, it would be like a rented warehouse for storing data above the clouds.
Files (such as texts and images) called objects are stored in a container called a bucket.
Data can be retrieved on an object-by-object basis, allowing flexible data storage as needed.
Summary
Static website hosting (can easily be made into a webserver)
S3 Select (query function that allows you to easily aggregate S3 data)
Easy to scale up and down
Features
Scalability
Availability and durability
Reliability
Extensive management functions
Intelligent features
First of all, S3 is object storage, which makes it easy to scale up and down.
Multiple storage classes are available for different usage scenarios and can be automatically migrated by using a lifecycle policy.
The ability to scale up and down according to usage and not having to worry about capacity limits is a major advantage of object storage.
S3 claims 99.99999999999% data durability, a feature that makes it resistant to threats such as failures and errors. S3 objects are automatically replicated and stored in at least three different availability zones, which allows users to continue using them even if one of them fails.
S3 also has encryption capabilities and access management tools to protect data from attacks.
Various management functions such as storage class analysis and life cycle policies are available in S3 and the management functions enable the selection of storage classes that fit the actual usage.
There are functions and services that perform data queries.
In addition, the amount of data that can be stored is unlimited.
Although there is a limit of up to 5 TB for a single file, there is no need to worry about the size limit since it is unlikely that you will access more than 5 TB of data frequently.
Pricing
There are five conditions under which fees are charged.
Storage
Requests and Data Retrieval
Data Transfer
Management and Analysis
Replication
Reference: Amazon S3 pricing
Conditions | Charging | Free Tier |
Storage | Charged for storage capacity. Rates vary by storage class. | Up to 5GB of standard storage class is available with a free S3 tier. |
Requests and Data Retrieval | Charges are made for the number of requests such as GET/PUT/POST/LIST/COPY/SELECT/DELETE etc. | Free monthly access to the following for one year. -20,000 GET requests, -2,000 PUT/COPY/POST/LIST requests -15GB of data transmission |
Data Transfer | Charged for transmission from S3. Charges are incurred when retrieving data stored in S3. Charges are incurred when data stored in S3 is retrieved from or transferred to another region. If you are using Transfer Acceleration, a Transfer Acceleration fee will be charged on top of the transfer fee. *Note that the fee varies depending on the region to which the data is transferred | Uploading to S3 is free. Transfers between S3 buckets or from S3 to other services within the same region are free. Transfers to Amazon CloudFront are free. But only for transfers from S3 to CloudFront. |
Management and Analysis | Charges will be incurred for storage management functions and analysis enabled in the account's bucket. | |
Replication | Charged for primary copy and replication PUT request storage fees. Low frequency access storage retrieval charges and S3 storage charges for the S3 storage class selected as the destination. | |
Storage Classes
S3 offers various types of storage that are referred to as storage classes.
Storage classes can be changed depending on the situation, and the change can also be done automatically by setting a lifecycle policy.
Storage Class | Features |
S3 Standard | Most standard storage class. Suitable for storing data such as images and videos. No charge for data retrieval. Data stored in three or more availability zones are guaranteed 99.9% availability. |
S3 Intelligent-Tiering | The two hierarchies, high-frequency access hierarchy and low-frequency access hierarchy, are moved according to the access frequency. The decision to store data in either the high-frequency or low-frequency hierarchy is monitored on an object-by-object basis, and the data is automatically moved according to the results of the monitoring. Objects that have not been accessed for 30 consecutive days are automatically moved to the low-frequency hierarchy. |
S3 Standard-IA | Storage costs are lower than standard storage, but data retrieval costs are higher. Data is stored in three or more availability zones, which works as disaster protection. Even if one availability zone is lost entirely, the data can be recovered. Although access frequency is low, it is suitable for long-term storage such as backups that require immediate data retrieval when data is needed. |
S3 One Zone-IA | Compared to the standard class, storage fees are lower, but transfer fees are higher. It is suitable for cases where access frequency is low but storage volume is high. However, since the data is stored in only one availability zone, there is a possibility of data loss if physical trouble occurs in that area. Therefore, it is not suitable for storing data that must not be lost despite its low cost. |
S3 Glacier | Suitable for data archiving and long-term backup. Large amounts of data can be stored at low cost. In the case of Glacier, data is stored in a container called a "vault". Therefore, when retrieving stored data, retrieval to another S3 bucket is required, which takes time and increases the transfer volume. Also, note that when data is retrieved, storage fees are charged for both the data on Glacier and the data at the retrieval destination. |
S3 Glacier Deep Archive | Lower cost than Glacier. Useful when data is rarely accessed but cannot be deleted. However, the cost of retrieving data is high and it takes a considerable amount of time to retrieve data. Therefore, it is suitable for long-term storage of data that is rarely accessed and that does not have to be accessed immediately when the data is needed. |
Summary
S3 is like a rented warehouse for storing data above the clouds, so if data is stored in a rented warehouse in a convenient location, the data can be retrieved quickly, but the better location makes it more expensive to store.
If data is stored in a warehouse in a distant location like Glacier, it can be stored at a lower cost due to its poor location although it takes time and effort (cost) to retrieve it because it is far to get to the warehouse.
With this in mind, you may be able to decide which storage class is best suited for your operation.
This blog post is translated from a blog post by Haruka Inoue on Beyond Co..