Amazon Timestream Cheat Sheet

• A fully managed, purpose-built time series database service for storing and analyzing trillions of time series data points per day
  • Offers multiple database engines: Timestream for InfluxDB (Recommended) – Managed InfluxDB databases for real-time time series applications with single-digit millisecond query response times and up to 99.9% availability
  • Timestream for InfluxDB 3 – Next-generation managed InfluxDB 3 databases leveraging Apache Arrow, Apache Data Fusion, and columnar Parquet storage on Amazon S3
  • Timestream for LiveAnalytics – Serverless, for large-scale data ingestion and SQL queries (No longer accepting new customers as of June 20, 2025 – existing customers can continue using the service)

Amazon Timestream for InfluxDB

• AWS’s recommended time series database for new customers
• Managed InfluxDB 2.x databases on AWS
• Compatible with open-source InfluxDB APIs, tools, and Telegraf plugins
  • DB Instance: Basic building block
  • Supports multiple organizations and buckets
  • Up to 40 instances per account
  • Instance Classes: db.influx.medium to db.influx.24xlarge
  • vCPU range: 1 to 96
  • Memory range: 8 GiB to 768 GiB

• Storage: Influx IOPS Included with three tiers (3K, 12K, 16K IOPS)
  • Deployments: Single-AZ
  • Multi-AZ (primary + standby in different AZ)
  • Read Replica Cluster (licensed add-on via AWS Marketplace)
• Authentication: Master user account, InfluxDB API tokens, AWS Secrets Manager integration

Amazon Timestream for InfluxDB 3

• Next-generation InfluxDB engine with a completely new architecture
• Built on Apache Arrow (in-memory processing), Apache Data Fusion (query execution), and Parquet columnar storage on Amazon S3
  • Deployment Options: Core – Single-node, no compaction, best for near real-time workloads on recent data
  • Enterprise – Multi-node clusters, compaction support, horizontal scaling, suitable for long-term analytics
• Storage: S3-based object storage, shared across all nodes, virtually unlimited capacity
• Query APIs: SQL, InfluxQL, HTTP, Flight+gRPC, v1 compatibility API
• Features: Last Value Cache (LVC) for sub-10ms response times

Amazon Timestream for LiveAnalytics

• Effective June 20, 2025: Amazon Timestream for LiveAnalytics is closed to new customers
• Existing customers: Can continue using the service typically; AWS continues investing in security, availability, and performance improvements.
• Recommendation: AWS recommends evaluating Amazon Timestream for InfluxDB as an alternative for new workloads
• Migration: AWS provides migration guides for transitioning from LiveAnalytics to InfluxDB
• Serverless, fully managed time series database that automatically scales up or down to adjust capacity and performance
• Ingest tens of gigabytes of time-series data per minute and run SQL queries on terabytes of data in seconds.

Amazon Timestream Integrations

• Data Ingestion: AWS IoT Core (rule actions)
  • Amazon Kinesis
  • Amazon MSK (Managed Streaming for Apache Kafka)
  • Open-source Telegraf
  • Apache Flink adapter
• Visualization:
  • Amazon QuickSight
  • Grafana (Timestream plugin)
  • Business intelligence tools via JDBC driver
• Machine Learning:
  • Amazon SageMaker AI
• Backup:
  • AWS Backup integration for LiveAnalytics (immutable backups, cross-account/Region copies)

Amazon Timestream Security

• Encryption: All data is encrypted at rest and in transit by default
  • Uses AWS KMS for key management
  • Memory store: Timestream service key
  • Magnetic store: Customer-managed KMS key (CMK) option
  • AES-256 encryption algorithm
• Access Control: AWS IAM integration for authentication and authorization
  • Identity-based policies for fine-grained access control
  • Resource-based tags for access control
• Network: VPC endpoints support
  • TLS 1.2 or later recommended
  • Perfect forward secrecy (PFS) cipher suites are required
• Compliance: Data replicated across 3 AZs for durability

Amazon Timestream Pricing

InfluxDB 2.x (Recommended)

• Compute – Per DB instance-hour (billed per second, 10-minute minimum)
• Storage – Per GB-month based on storage type and allocated volume

InfluxDB 3

• Compute – Per node instance-hour
• Storage – Per GB-month (S3-based object storage)

LiveAnalytics (Existing Customers Only)

• Writes – Per million writes of 1 KiB
• Memory Store – Per GB-hour
• Magnetic Store – Per GB-month (minimum 100 GB per account per Region)
• Queries – Per TCU-hour

Amazon Timestream Use Cases

• DevOps monitoring (infrastructure performance metrics)
• IoT sensor data storage and analysis
• Industrial telemetry for equipment management and maintenance
• Real-time application monitoring and alerting
• User behavior analytics and clickstream data
• Fraud detection in payment processing
• Fleet management and logistics optimization

Amazon Timestream Best Practices

• Data Modeling: Use multi-measure records for cost efficiency
  • Choose appropriate dimensions for time series identification.
  • Set retention policies based on query patterns and cost requirements.
• Writes: Batch writes (5000 lines per request for InfluxDB, 100 records for LiveAnalytics)
  • Sort tags by key lexicographically
  • Use the coarsest time precision possible.
  • Enable gzip compression
• Queries: Use scheduled queries for frequently accessed aggregates
  • Include time ranges, measure names, and dimension names in predicates for efficient pruning.
  • Monitor QueryTCU metric to optimize costs.
• Storage: Set memory store retention based on late-arriving data requirements
  • Consider magnetic store for long-term analytical queries.
  • Enable magnetic store writes for late-arriving data.

Amazon Timestream References:

• https://docs.aws.amazon.com/timestream/latest/developerguide/
• https://docs.aws.amazon.com/timestream/latest/developerguide/AmazonTimestreamForLiveAnalytics-availability-change.html
• https://aws.amazon.com/timestream/
• https://aws.amazon.com/timestream/pricing/
• https://aws.amazon.com/timestream/features/
• https://aws.amazon.com/products/lifecycle/