Timestream

🧠 What is Amazon Timestream?

Amazon Timestream is a fully managed, purpose-built time series database optimized for data such as application metrics, IoT sensor data, and DevOps monitoring logs.

Feature	Description
Time series optimized	Stores data with time-based partitions, ingestion, and compression
Serverless	No servers or capacity planning required
Multi-tier storage	Data moves from memory → magnetic automatically (configurable retention)
Fast queries	Optimized query engine for time series analytics (aggregations, filters)
SQL-based	Uses Timestream Query Language (subset of SQL)

---

📊 Use Cases

• Application performance monitoring (APM)

• Infrastructure and system metrics (CPU, memory, latency)

• IoT sensor data (temperature, humidity, GPS)

• Business KPIs over time

• Event-driven architecture metrics

---

⚙️ Architecture Overview

                    +----------------------------+
     Data Sources → |   Amazon Timestream       | ← Queries via SDK / CLI / Console
                    +----------------------------+
                           ↓            ↓
              Memory Store       Magnetic Store
              (fast access)       (cheap, historical)

---

🔧 Key Concepts

Concept	Description
Database	Logical group of tables
Table	Stores time series data in dimensions, measures, and time
Measure	The numeric or string value being recorded (e.g., temperature)
Dimension	Metadata that describes the source (e.g., device_id, region)
Time	The timestamp of the measurement
Memory store	Fast, in-memory, short-term storage
Magnetic store	Long-term, cost-efficient storage

---

🏗️ Creating a Timestream Database and Table

Step 1: Create a Database

Using CLI

aws timestream-write create-database \
  --database-name MyIoTDB

---

Step 2: Create a Table

aws timestream-write create-table \
  --database-name MyIoTDB \
  --table-name SensorMetrics \
  --retention-properties MemoryStoreRetentionPeriodInHours=24,MagneticStoreRetentionPeriodInDays=7

• Data stays in memory for 24 hours, then moves to magnetic storage for 7 days.

---

📝 Sample Record Format

A Timestream record contains:

{
  "Dimensions": [
    { "Name": "device_id", "Value": "sensor-001" },
    { "Name": "location", "Value": "factory-1" }
  ],
  "MeasureName": "temperature",
  "MeasureValue": "31.2",
  "MeasureValueType": "DOUBLE",
  "Time": "1718203893000"
}

---

🚀 Writing Data to Timestream

Using AWS SDK (Python Example)

import boto3
import time

client = boto3.client('timestream-write')

response = client.write_records(
    DatabaseName='MyIoTDB',
    TableName='SensorMetrics',
    Records=[
        {
            'Dimensions': [
                {'Name': 'device_id', 'Value': 'sensor-001'},
                {'Name': 'location', 'Value': 'factory-1'}
            ],
            'MeasureName': 'temperature',
            'MeasureValue': '30.5',
            'MeasureValueType': 'DOUBLE',
            'Time': str(int(time.time() * 1000))  # Unix epoch in ms
        }
    ]
)

---

🔍 Querying Data

Timestream uses SQL-compatible syntax to analyze time series data.

Sample Queries

-- Get last 10 temperature records
SELECT * FROM "MyIoTDB"."SensorMetrics"
WHERE MeasureName = 'temperature'
ORDER BY time DESC
LIMIT 10;

-- Average temperature per device in last 1 hour
SELECT device_id,
       AVG(measure_value::double) AS avg_temp
FROM "MyIoTDB"."SensorMetrics"
WHERE MeasureName = 'temperature'
  AND time > ago(1h)
GROUP BY device_id;

-- Count all metrics per 5-minute window
SELECT bin(time, 5m) AS binned_time,
       COUNT(*) as records
FROM "MyIoTDB"."SensorMetrics"
WHERE time > ago(12h)
GROUP BY bin(time, 5m)
ORDER BY binned_time DESC;

---

📈 Visualization

Tool	Integration
Amazon QuickSight	Native connector
Grafana	Timestream plugin available
AWS IoT Core	Can directly write to Timestream
Lambda	Stream data into Timestream

---

🔐 Security

Feature	Description
IAM Auth	Fine-grained write & query permissions
Encryption	KMS-managed encryption at rest
Private VPC Access	Supported via Interface VPC Endpoints
CloudTrail	Logs all access and usage

---

🔔 Monitoring and Logging

Service	What it Monitors
CloudWatch	Write success/failure metrics
CloudTrail	Logs API activity
Service Quotas	Limits on databases, tables, write rate

---

💰 Pricing

Cost Component	Approx. Cost
Write Requests	$0.50 per million write units
Query	$0.01 per GB scanned
Memory Store	$0.036 per GB/hour
Magnetic Store	$0.03 per GB/month

Pro Tip: Tune retention settings to reduce memory storage costs.

---

🧰 Terraform Sample

resource "aws_timestreamwrite_database" "iot" {
  database_name = "MyIoTDB"
}

resource "aws_timestreamwrite_table" "metrics" {
  database_name = aws_timestreamwrite_database.iot.database_name
  table_name    = "SensorMetrics"

  retention_properties {
    memory_store_retention_period_in_hours = 24
    magnetic_store_retention_period_in_days = 7
  }
}

---

✅ Summary

Feature	Description
Purpose	Time series data ingestion and analysis
Storage Tier	Memory (short-term), Magnetic (long-term)
Query Language	SQL-compatible (Timestream SQL)
Write Format	Dimensions + Measure + Time
Use Case	IoT, app monitoring, business metrics
Security	IAM, encryption, VPC endpoints
Pricing	Pay-per-use (write/query/storage)

---