📊 Redshift Serverless

Master Redshift Serverless auto-scaling, cost model, and best practices.

Module: AWS Data Engineering • Topic 38 of 65 • Premium Content

Serverless Architecture

Architecture Diagram

┌─────────────────────────────────────────────────────────────────────────────┐
│                    REDSHIFT SERVERLESS ARCHITECTURE                           │
│                                                                             │
│  ┌─────────────────────────────────────────────────────────────────────┐    │
│  │  WORKGROUP CONFIGURATION                                             │    │
│  │  Base Capacity: 128 - 512 RPUs                                      │    │
│  │  Auto-pause: 30 sec - 60 min                                        │    │
│  │  Managed Storage: S3-backed, auto-scaling                           │    │
│  └─────────────────────────────┬───────────────────────────────────────┘    │
│                                │                                           │
│              ┌─────────────────┼─────────────────┐                         │
│              ▼                 ▼                 ▼                         │
│  ┌─────────────────┐  ┌─────────────────┐  ┌─────────────────┐            │
│  │  Query Load     │  │  Auto-scaling   │  │  Auto-pause     │            │
│  │  128 → 256 →   │  │  RPUs based on  │  │  When idle      │            │
│  │  512 RPUs      │  │  query demand   │  │  (configurable) │            │
│  └─────────────────┘  └─────────────────┘  └─────────────────┘            │
│                                                                             │
│  COST MODEL:                                                                │
│  RPU-hour = RPU count × hours used                                         │
│  128 RPUs × 1 hour = $12.80                                               │
│  256 RPUs × 1 hour = $25.60                                               │
│  Storage: $0.024/GB/month                                                  │
└─────────────────────────────────────────────────────────────────────────────┘

Serverless Configuration

import boto3

redshift = boto3.client('redshift-serverless')

# Create workgroup
response = redshift.create_workgroup(
    workgroupName='analytics-serverless',
    baseCapacity=128,
    enhancedVpcRouting=True,
    securityGroupIds=['sg-12345678'],
    subnetIds=['subnet-12345678', 'subnet-87654321'],
    publiclyAccessible=False,
    configParameters=[
        {'parameterKey': 'enable_user_activity_logging', 'parameterValue': 'true'},
        {'parameterKey': 'max_query_execution_time', 'parameterValue': '3600'}
    ]
)

# Create namespace
namespace = redshift.create_namespace(
    namespaceName='analytics-namespace',
    adminUsername='admin',
    adminUserPassword='SecurePassword123!',
    dbName='analytics_db',
    iamRoles=['arn:aws:iam::123456789012:role/ServerlessRole']
)

Interview Q&A

Q1: Serverless vs Provisioned?

Answer: Serverless for variable workloads, dev/test. Provisioned for steady-state production with predictable costs.

Q2: How does auto-scaling work?

Answer: RPUs scale automatically based on query complexity. Base capacity sets minimum; max is 4x base or 512 RPUs.

Q3: What is the minimum auto-pause setting?

Answer: 5 minutes. Redshift Serverless will pause after 5 minutes of inactivity to save costs.

Summary

Architecture: Serverless auto-scaling based on query demand
Pricing: Per RPU-hour + managed storage
Auto-pause: Configurable 5 min to 60 min
Best For: Variable workloads, dev/test, unpredictable queries
Storage: S3-backed managed storage, auto-scaling

Redshift Serverless Deep Dive