SQLBits 2026
Autonomous DBOps: Agentic AI for Maintaining SQL Server Databases
We will demonstrate about a self-healing database systems using AI agents powered by AWS Strands SDK. Learn to create autonomous agents that diagnose performance bottlenecks, generate optimizations, and implement fixes—transforming reactive database operations into proactive, intelligent automation.
Imagine, It's 2 AM on a Tuesday. Your database cluster, the beating heart of your e-commerce platform is suddenly choking. CPU is pegged at 95%, queries are timing out, and you're on-call to fix the issue.
Here's the challenge: you have exactly 30 minutes to identify the culprits, tune the database, and restore stability before leadership wakes up demanding answers.
Sound familiar? We've all been there—manually sifting through CloudWatch metrics, tweaking parameters in panic mode, and hoping our interventions don't make things worse. But what if you didn't have to? What if you had an autonomous Database Agent, a reasoning sidekick, that could diagnose bottlenecks, hypothesize fixes, and implement them autonomously, all while explaining its logic like a senior DBA?
In this session, we're diving headfirst into that world. You'll see a running Amazon RDS for SQL Server database experiencing real performance issues. Learn how to restore the database to healthy performance by building a self-healing system with AI agents powered by the AWS Strands.
Health Check Agent — Identifies top queries by execution time, retrieves execution plans, detects index fragmentation, identifies missing and unused indexes, and generates actionable recommendations
Actions Agent — Implements database optimizations including creating indexes, updating statistics, rebuilding fragmented indexes, and optimizing query plans
Supervisor Agent — Coordinates both agents to form an autonomous system that identifies issues, proposes fixes, and applies them with minimal production impact
CloudWatch Integration — Deploy the Health Check Agent to Amazon Bedrock AgentCore runtime. When triggered by CloudWatch alarms (e.g., CPU exceeding 80%), the agent automatically diagnoses issues and emails recommendations to operators
AWS Services Used:
Amazon RDS for SQL Server
Amazon CloudWatch Database Insights
Amazon Bedrock AgentCore
AWS Strands SDK
Here's the challenge: you have exactly 30 minutes to identify the culprits, tune the database, and restore stability before leadership wakes up demanding answers.
Sound familiar? We've all been there—manually sifting through CloudWatch metrics, tweaking parameters in panic mode, and hoping our interventions don't make things worse. But what if you didn't have to? What if you had an autonomous Database Agent, a reasoning sidekick, that could diagnose bottlenecks, hypothesize fixes, and implement them autonomously, all while explaining its logic like a senior DBA?
In this session, we're diving headfirst into that world. You'll see a running Amazon RDS for SQL Server database experiencing real performance issues. Learn how to restore the database to healthy performance by building a self-healing system with AI agents powered by the AWS Strands.
Health Check Agent — Identifies top queries by execution time, retrieves execution plans, detects index fragmentation, identifies missing and unused indexes, and generates actionable recommendations
Actions Agent — Implements database optimizations including creating indexes, updating statistics, rebuilding fragmented indexes, and optimizing query plans
Supervisor Agent — Coordinates both agents to form an autonomous system that identifies issues, proposes fixes, and applies them with minimal production impact
CloudWatch Integration — Deploy the Health Check Agent to Amazon Bedrock AgentCore runtime. When triggered by CloudWatch alarms (e.g., CPU exceeding 80%), the agent automatically diagnoses issues and emails recommendations to operators
AWS Services Used:
Amazon RDS for SQL Server
Amazon CloudWatch Database Insights
Amazon Bedrock AgentCore
AWS Strands SDK