DynamoDB

DynamoDB is a fully managed NoSQL database service offering consistent, single-digit millisecond latency at any scale. It supports both document and key-value data models, making it ideal for mobile, web, gaming, ad-tech, IoT, and many other applications.

Key Points

1. Switching between modes:

   • Can switch between Provisioned and On-Demand once per 24 hours

2. Global Tables:

   • Both modes support DynamoDB Global Tables for multi-region replication

3. Backup and Restore:

   • Both modes support on-demand backup and restore

4. Read Consistency:

   • Both modes support eventually consistent and strongly consistent reads

5. DAX (DynamoDB Accelerator):

   • Compatible with both capacity modes

6. Use case identification:

   • Be prepared to recommend the appropriate capacity mode based on given scenarios

Secondary Indexes

In DynamoDB, you can have:

• Up to 20 Global Secondary Indexes (GSIs) per table

• Up to 5 Local Secondary Indexes (LSIs) per table

Note that LSIs must be created when the table is created, while GSIs can be added or removed at any time.

Core Features

Storage and Consistency

• Stored on SSD storage

• Data spread across 3 geographically distinct Availability Zones (AZs) withign the same Region.

• Two consistency models:

  • Eventually consistent reads (default): Updates typically propagate within one second

  • Strongly consistent reads: Returns results reflecting all successful writes prior to the read

DynamoDB Accelerator (DAX)

• Fully managed, highly available, in-memory cache

• Improves performance by 10x

• Reduces request time from milliseconds to microseconds

• Automatically handles caching logic

• Compatible with DynamoDB API calls

Capacity Modes Options

DynamoDB has three capacity modes for managing throughput:

1. Provisioned Capacity (default)

• You specify Read Capacity Units (RCU) and Write Capacity Units (WCU)

• Pay for provisioned capacity whether used or not

• Can enable Auto Scaling

• Good for predictable workloads

Capacity Units Calculation

• Use Case:

  • For predictable workloads

  • When you can forecast read and write capacity requirements

• How it works:

  You specify Read Capacity Units (RCUs) and Write Capacity Units (WCUs)

  • 1 RCU = 1 strongly consistent read/sec for items up to 4KB

  • 1 RCU = 2 eventually consistent reads/sec for items up to 4KB

  • 1 WCU = 1 write/sec for items up to 1KB

• Scaling:

  • Can use Auto Scaling to automatically adjust provisioned capacity

  • Set upper and lower bounds for scaling

• Cost:

  • Most cost-effective for predictable workloads

  • Pay for provisioned capacity whether you use it or not

• Performance:

  • Consistent performance if capacity is set correctly

  • Can lead to throttling if capacity is set too low

• Effort:

  • Requires monitoring and adjusting capacity based on usage patterns

  • Need to review past usage to set appropriate scaling bounds

2. On-Demand Capacity

• Pay-per-request pricing

• No capacity planning needed

• More expensive per request

• Good for unpredictable workloads

• No throttling

• Use Case:

  • For unpredictable workloads

  • New applications with unknown traffic patterns

  • Applications with sporadic, short-lived spikes in traffic

• How it works:

  • Automatically scales up and down based on actual traffic

  • No need to specify RCUs and WCUs

• Scaling:

  • Instantaneous and unlimited scaling

  • Can handle sudden and large spikes in traffic

• Cost:

  • Pay per request pricing

  • Generally more expensive than well-optimized Provisioned capacity

  • No charges for read and write capacity that you don't use

• Performance:

  • Consistent performance regardless of load

  • No throttling due to capacity misconfigurations

• Effort:

  • Simple to use - just select On-Demand mode

  • No capacity planning required

3. Global Tables

• Multi-region write capability

• Uses On-Demand or Provisioned

• Active-active configuration

• Sub-second latency

DynamoDB limits and constraints

Item Size Limits:

• Maximum item size: 400 KB (including attribute names and values)

Table Limits:

• No limit on number of items per table

• No limit on table size

• Maximum number of tables per account: 2500 by default (can be increased)

• Maximum partition key length: 2048 bytes

• Maximum sort key length: 1024 bytes

Throughput Limits (for provisioned capacity mode):

• Per table minimum: 1 RCU and 1 WCU

• Per table maximum: 40,000 RCU and 40,000 WCU (can be increased)

• Per partition: 3,000 RCU and 1,000 WCU

Secondary Indexes:

• Maximum number of local secondary indexes per table: 5

• Maximum number of global secondary indexes per table: 20

• Each secondary index adds another copy of data with different key schema

API Specific Limits:

• BatchGetItem: Up to 100 items or 16 MB per request

• BatchWriteItem: Up to 25 items or 16 MB per request

• Query and Scan: Result set size limit of 1 MB per call

• Transactional operations: Up to 25 items or 4 MB per transaction

Attribute Naming and Data Types:

• Attribute names: Maximum length of 64 KB

• Number of attributes per item: No fixed limit, but must fit within 400 KB item size

• Supports data types: String, Number, Binary, Boolean, Null, Lists, Maps, Sets

Time To Live (TTL):

• One TTL attribute per table

• Must be a Number data type representing epoch timestamp

Auto Scaling Settings

• Target utilization (default 70%)

• Minimum capacity

• Maximum capacity

Security Features

• KMS encryption at rest

• Site-to-site VPN support

• Direct Connect (DX) integration

• IAM policies and roles

• Fine-grained access control

• CloudWatch and CloudTrail integration

• VPC endpoints

Advanced Features

DynamoDB Transactions

Supports ACID transactions across multiple tables within the same AWS account and region:

• Atomic: All changes must succeed or none are applied

• Consistent: Data remains consistent before and after transactions

• Isolated: Transactions are protected from interference

• Durable: Changes persist after completion

Transaction options:

• Read options:

  • eventual consistency (default): When reading data, you might get a response that doesn't reflect the results of a recently completed write

  • strong consistency: Provides consistency for a single item operation. Doesn't provide isolation from concurrent modifications

  • transactional: Provides consistency across multiple items and tables (ACID transaction)

• Write options: standard or transactional

• Limits: Up to 25 items or 4 MB of data

Backup and Recovery

On-Demand Backup:

• Full backups available anytime

• No performance impact

• Consistent within seconds

• Same-region operation

Point-in-Time Recovery:

• 35-day recovery window

• Incremental backup system

• 5-minute recovery point objective (RPO)

• Optional feature

DynamoDB Streams

A Stream is divided into Shards and a shard contains Stream Data.

• Time-ordered sequence of item-level modifications

• 24-hour retention period

• Captures inserts, updates, and deletes

• Lambda function integration for stored procedure-like functionality

Global Tables

• Multi-master, multi-region replication

• Sub-second replication latency

• Built on DynamoDB streams

• Requires streams enablement

• No application modifications needed

• Replication latency is under 1 second

Important Considerations

Technical Aspects

• Partition management is automatic but requires understanding of partition keys

• Supports Global Secondary Indexes (GSI) and Local Secondary Indexes (LSI)

• Read/Write Capacity Units (RCU/WCU) calculations vary by item size and consistency

• Auto-scaling available for both provisioned and on-demand capacity

Integration and Optimization

• Seamless integration with AWS services

• Cost optimization through capacity mode selection, DAX usage, and TTL

• Key limits: 400 KB maximum item size, 2 KB maximum partition key size

• Common patterns include Lambda integration and session state management