TTL behavior and timing in DynamoDB - Deep Dive

Overview - TTL behavior and timing

What is it?

TTL stands for Time To Live. It is a feature in DynamoDB that automatically deletes items after a specified time. You set a timestamp attribute on each item, and when the current time passes that timestamp, DynamoDB removes the item. This helps keep your database clean without manual deletion.

Why it matters

Without TTL, expired or outdated data would pile up, wasting storage and slowing down queries. Manually deleting old data is error-prone and costly. TTL automates cleanup, saving money and improving performance. It also helps keep data relevant and compliant with retention policies.

Where it fits

Before learning TTL, you should understand basic DynamoDB concepts like tables, items, and attributes. After TTL, you can explore advanced data lifecycle management, backup strategies, and cost optimization in DynamoDB.

Mental Model

Core Idea

TTL is like a self-destruct timer on each database item that triggers automatic deletion when time runs out.

Think of it like...

Imagine a library book with a due date sticker. When the due date passes, the book is automatically removed from the shelf without needing a librarian to check it out manually.

┌───────────────┐
│   DynamoDB    │
│   Table       │
│ ┌───────────┐ │
│ │ Item 1    │ │
│ │ TTL=165000│ │
│ └───────────┘ │
│ ┌───────────┐ │
│ │ Item 2    │ │
│ │ TTL=165100│ │
│ └───────────┘ │
└─────┬─────────┘
      │
      ▼
  Time passes → TTL expires → Item deleted automatically

Build-Up - 7 Steps

1

Foundation

What is TTL in DynamoDB

Concept: Introduce the basic idea of TTL as a timestamp attribute that triggers automatic deletion.

TTL is a special attribute you add to your DynamoDB items. It stores a Unix epoch time (seconds since 1970). When the current time passes this value, DynamoDB marks the item for deletion. You enable TTL on a table and specify which attribute holds the TTL timestamp.

Result

Items with expired TTL values are automatically deleted without manual intervention.

Understanding TTL as a timestamp attribute helps you see how DynamoDB knows when to delete items automatically.

2

Foundation

How to enable TTL on a table

Concept: Explain the setup process to activate TTL on a DynamoDB table.

To use TTL, you must enable it on your DynamoDB table and specify the attribute name that holds the TTL timestamp. This is done via the AWS Console, CLI, or SDK. Once enabled, DynamoDB starts monitoring that attribute for expiration.

Result

TTL monitoring begins, and items with expired timestamps will be deleted automatically.

Knowing that TTL must be explicitly enabled and configured prevents confusion about why items are not deleted automatically.

3

Intermediate

How TTL deletion timing works

🤔Before reading on: Do you think DynamoDB deletes expired items immediately when TTL expires, or is there a delay? Commit to your answer.

Concept: Explain the approximate timing and process DynamoDB uses to delete expired items.

DynamoDB does not delete expired items instantly at the TTL timestamp. Instead, it runs a background process that scans for expired items periodically, usually within 48 hours after expiration. This means there can be a delay between expiration and actual deletion.

Result

Expired items remain visible for some time after TTL passes, then get deleted automatically later.

Understanding the delay helps set realistic expectations and prevents surprises when expired data still appears briefly.

4

Intermediate

TTL attribute data requirements

🤔Before reading on: Do you think TTL attribute can be any data type, or must it be a number? Commit to your answer.

Concept: Clarify the data type and format requirements for the TTL attribute.

The TTL attribute must be a Number type representing Unix epoch time in seconds. If the attribute is missing or not a valid number, DynamoDB ignores TTL for that item. You cannot use strings or other data types for TTL.

Result

Only items with a valid numeric TTL attribute are considered for automatic deletion.

Knowing the strict data type requirement prevents errors and ensures TTL works as expected.

5

Intermediate

What happens after TTL deletion

Concept: Describe the state of the item and table after TTL removes expired items.

When DynamoDB deletes an item due to TTL, it removes it permanently from the table. This deletion is irreversible and does not trigger DynamoDB Streams or backups automatically. You must plan for data recovery or auditing separately.

Result

Expired items are gone from the table and cannot be retrieved unless you have backups.

Understanding the permanence of TTL deletion helps you design backup and audit strategies accordingly.

6

Advanced

TTL impact on billing and performance

🤔Before reading on: Do you think TTL deletion reduces your DynamoDB costs immediately, or is the effect delayed? Commit to your answer.

Concept: Explain how TTL affects storage costs and read/write capacity over time.

TTL reduces storage costs by removing expired items, but the cost savings appear only after deletion occurs, which can take up to 48 hours. TTL deletions do not consume write capacity units but may slightly increase read capacity during background scans. Proper TTL use can optimize costs and performance.

Result

Storage costs decrease gradually as expired items are deleted; read/write capacity usage remains mostly unaffected.

Knowing the delayed cost impact and capacity behavior helps optimize DynamoDB usage and budgeting.

7

Expert

TTL internal deletion mechanism and limitations

🤔Before reading on: Do you think TTL deletions trigger DynamoDB Streams or transactional events? Commit to your answer.

Concept: Reveal the internal process DynamoDB uses to delete expired items and its limitations.

DynamoDB's TTL process scans tables asynchronously and deletes expired items without generating DynamoDB Stream events or transactional logs. This means TTL deletions are invisible to triggers or change data capture. Also, TTL does not guarantee exact deletion time, and items may persist longer if the background process is delayed.

Result

TTL deletions happen silently and asynchronously, without event notifications or strict timing guarantees.

Understanding these internal details prevents incorrect assumptions about event-driven workflows and timing precision.

Under the Hood

DynamoDB stores the TTL attribute as a numeric timestamp. A background service periodically scans the table for items where the TTL timestamp is less than the current time. It then deletes those items asynchronously. This process runs independently from user operations and does not generate stream events or transactional logs.

Why designed this way?

The asynchronous design reduces load on the main database operations and avoids slowing down user queries or writes. Immediate deletion would require constant scanning and locking, which would hurt performance. The lack of stream events for TTL deletions simplifies internal complexity but limits event-driven use cases.

┌───────────────┐       ┌───────────────┐
│ User writes   │       │ Background    │
│ item with TTL │──────▶│ TTL scanner   │
│ attribute     │       │ runs periodically│
└───────────────┘       └───────┬───────┘
                                   │
                                   ▼
                          ┌─────────────────┐
                          │ Delete expired   │
                          │ items silently   │
                          └─────────────────┘

Myth Busters - 4 Common Misconceptions

Quick: Does TTL delete items exactly at the expiration second? Commit yes or no.

Common Belief: TTL deletes items immediately when the timestamp expires.

Reality: TTL deletions happen asynchronously and can be delayed up to 48 hours after expiration.

Why it matters: Expecting immediate deletion can cause confusion and errors in applications relying on strict data expiry.

Quick: Do TTL deletions trigger DynamoDB Streams events? Commit yes or no.

Common Belief: TTL deletions generate stream events like normal deletes.

Reality: TTL deletions do not trigger DynamoDB Streams or any event notifications.

Why it matters: Relying on streams to detect TTL deletions will fail, breaking event-driven workflows.

Quick: Can TTL attribute be a string representing a date? Commit yes or no.

Common Belief: TTL attribute can be any data type, including strings or ISO date formats.

Reality: TTL attribute must be a Number type representing Unix epoch time in seconds.

Why it matters: Using wrong data types causes TTL to be ignored, leaving expired items undeleted.

Quick: Does TTL deletion automatically back up expired items? Commit yes or no.

Common Belief: TTL deletions are backed up or archived automatically by DynamoDB.

Reality: TTL deletions permanently remove items without automatic backups.

Why it matters: Assuming backups exist can lead to data loss if recovery is needed.

Expert Zone

1

TTL deletions do not consume write capacity units but may cause slight read capacity usage due to background scans.

2

TTL attribute updates reset the expiration timer; changing the TTL value extends or shortens item lifetime.

3

TTL does not guarantee deletion order or exact timing, so expired items may coexist briefly with unexpired ones.

When NOT to use

TTL is not suitable when you need immediate deletion, event notifications on deletes, or transactional guarantees. Alternatives include manual deletion with DynamoDB Streams or using DynamoDB Transactions for precise control.

Production Patterns

In production, TTL is used to automatically expire session data, cache entries, or temporary logs. It is combined with backups and streams for auditing. Developers monitor TTL deletion delays and design applications to tolerate eventual consistency in data expiry.

Connections

Cache expiration

TTL in DynamoDB is similar to cache expiration policies in caching systems.

Understanding TTL helps grasp how caches automatically remove stale data to keep responses fresh.

Garbage collection (computer science)

TTL deletion is a form of garbage collection that cleans up unused data automatically.

Knowing garbage collection concepts clarifies why asynchronous cleanup improves system performance.

Legal data retention policies

TTL supports compliance by enforcing automatic data deletion after a retention period.

Recognizing TTL's role in legal compliance helps design systems that respect privacy and regulations.

Common Pitfalls

#1 Setting TTL attribute as a string instead of a number

Wrong approach:

Item example: { "id": "123", "expireAt": "2024-06-01T00:00:00Z" }

Correct approach:

Item example: { "id": "123", "expireAt": 1711929600 } // Unix epoch seconds

Root cause:

Misunderstanding TTL attribute data type requirements causes DynamoDB to ignore TTL.

#2 Expecting immediate deletion at TTL expiration

Wrong approach:

Application logic assumes item is gone right after TTL timestamp passes.

Correct approach:

Design application to tolerate up to 48 hours delay before item is deleted.

Root cause:

Not knowing TTL deletion is asynchronous leads to incorrect assumptions about data availability.

#3 Relying on DynamoDB Streams to detect TTL deletions

Wrong approach:

Triggering workflows based on stream events for TTL-deleted items.

Correct approach:

Use alternative methods like periodic scans or application logic to handle expired items.

Root cause:

Incorrect belief that TTL deletions generate stream events causes broken event-driven designs.

Key Takeaways

TTL in DynamoDB automatically deletes items after a specified timestamp to keep data fresh and save costs.

TTL deletions happen asynchronously and can be delayed up to 48 hours after expiration, not instantly.

The TTL attribute must be a numeric Unix epoch time in seconds; wrong data types cause TTL to be ignored.

TTL deletions do not trigger DynamoDB Streams or backups, so plan accordingly for auditing and recovery.

Understanding TTL's behavior and limitations helps design reliable, cost-effective, and compliant DynamoDB applications.