AWS TL;DR: Amazon S3 Glacier (Archive Storage)
Back to AWS TL;DR Hub

Amazon S3 Glacier

/tldr: Extremely low-cost, durable, and highly available storage for data archiving and long-term backup.

Archive Storage Extremely Low Cost Minutes/Hours Retrieval

1. The Core Purpose: Deep Archive

S3 Glacier is designed for data that is infrequently accessed, often once a quarter or less, and where a few hours of retrieval time is acceptable. It is the cheapest and most durable storage service AWS offers, making it ideal for regulatory compliance, medical records, media archives, and data retention mandates.

Key Characteristics

  • **Durability:** 99.999999999% (11 nines) durability across multiple AZs.
  • **Cost:** Storage cost is measured in fractions of a penny per GB/month (significantly lower than S3 Standard).
  • **Retrieval:** The primary trade-off is retrieval time and cost, which are explicitly part of the billing model.

2. The Three Glacier Classes

Glacier is now a family of storage classes within the S3 service, each optimized for different access patterns.

S3 Glacier Instant Retrieval

Purpose: Data that needs immediate retrieval but is accessed rarely (once a quarter).

  • **Access Time:** Milliseconds (like S3 Standard).
  • **Cost:** Cheaper storage than S3 Standard-IA, but higher retrieval cost.
  • **Minimum Duration:** 90 days.

S3 Glacier Flexible Retrieval

Purpose: Long-term archives where data retrieval is acceptable in minutes or hours.

  • **Access Time:** Minutes to hours (requires explicit retrieval request).
  • **Cost:** Cheaper than Instant Retrieval.
  • **Minimum Duration:** 90 days.

S3 Glacier Deep Archive

Purpose: The lowest-cost storage for data retained for 7–10 years or more (e.g., tape replacement).

  • **Access Time:** 12 to 48 hours (requires explicit retrieval request).
  • **Cost:** Lowest storage cost of all S3 classes.
  • **Minimum Duration:** 180 days.

3. Retrieval Tiers (The Core Trade-off)

For **Flexible Retrieval** and **Deep Archive** classes, you must choose a retrieval tier when requesting data. The tier determines how quickly the data becomes available (it is temporarily moved to the S3 Standard class) and how much you pay.

// TIER 1: Expedited Retrieval (Highest Cost)
// Time: 1–5 minutes
// Used for: Urgent access to a subset of archives.

// TIER 2: Standard Retrieval (Moderate Cost)
// Time: 3–5 hours (Flexible) | 12 hours (Deep Archive)
// Used for: Most common retrieval needs.

// TIER 3: Bulk Retrieval (Lowest Cost)
// Time: 5–12 hours (Flexible) | 48 hours (Deep Archive)
// Used for: Retrieving large amounts of data where time is not critical.
            

How to Store Data in Glacier

Data can be stored directly in the Glacier classes, but the most common practice is using S3 Lifecycle Rules:

  • Transition: Define a rule to automatically move data from S3 Standard (or S3-IA) to a Glacier class after a certain number of days (e.g., move files older than 60 days to Glacier Flexible).

Glacier is for cold storage. You save money on storage, but pay time and money to get it back.

AWS Fundamentals Series: Amazon S3 Glacier