Database O&M often relies on passive firefighting. CPU usage spikes, deadlocks happen, and slow queries disrupt services. Database administrators (DBAs) can only react. This is a real pain point for many technical teams.
If the CPU usage of an online database suddenly spikes to 99.76%, what do you do?
You check monitoring, read the slow query log, find the problematic SQL, assess the impact, and apply rate limiting.
This flow is logically sound but too slow for the time window. It takes 10 minutes to several hours. Business loss often occurs in these critical few minutes.
DAS Agent provided by ApsaraDB team offers another answer.
DAS Agent is an intelligent database O&M brain. It combines Large Language Model (LLM) technology with expert experience from 100,000 Alibaba Cloud tickets. It builds an autonomous system covering issue discovery, root cause analysis, and closed-loop optimization. This end-to-end autonomous system keeps core database services online 24/7. It helps thousands of enterprises enter the AI-native O&M era.
Consider a typical scenario where CPU usage spikes to 99%.
• Autonomous anomaly discovery: It detects that CPU usage reaches 99.76% and pinpoints the problematic SQL.
• Provide optimization paths: It provides index suggestions, SQL rewrites, and immediate remediation measures in one step.
• Initiate rate limiting directly: You complete the operation in the dialog box without navigating to other consoles.
• Deliver immediate results: After rate limiting, CPU usage drops from 99.76% to 1.05%.
Next, DAS Agent independently completes a deadlock analysis. It does more than read logs and reports. It checks whether the instance has auditing enabled. It retrieves the full SQL log. It autonomously decides the query time range. Finally, it renders an accurate transaction time series chart for the deadlock.
The entire procedure requires no manual intervention or preset scripts. The AI autonomously decides what to query, which API to invoke, and how to combine skills at each step.
Some may ask: We use a self-managed MySQL database that is not on Alibaba Cloud. Can DAS Agent still provide O&M services?
The answer is: Yes!
DAS Agent supports host or centralized gateway patterns. It connects to MySQL, PostgreSQL, Redis, and MongoDB instances that are self-managed or on other clouds. This requires no data migration.
In the video, after DAS Agent connects to a self-managed MySQL instance:
1. It uses the IP address and port to autonomously query the instance ID. It turns pages multiple times, retries failed operations, and automatically adjusts the plan. No one tells it what to do.
2. Storage analysis: The instance has used 48 GB. The usage rate is 81.4%. It is expected to last for another 99 days. It clearly renders the database and table schema and the top large tables.
3. SQL dry run: You send an SQL statement to be published to it. DAS Agent invokes the execution plan, table schema, and engine information to provide three specific issues:
• The GROUP BY field is not the leftmost prefix of the index. You must create a temporary table.
• ORDER BY triggers additional file sorting. The existing index is not used.
• The join of two tables has low efficiency because filter conditions are unevenly distributed.
4. Provide optimization suggestions: Creation of a composite index and SQL rewriting (filter before join). These are directly executable.
Some enterprises now integrate this capability into their publishing flow. Audit logs automatically catch new SQL statements in the staging environment. These statements go to DAS Agent for a dry run based on the production environment. This requires no actual running. This blocks issues before they reach the production environment.
Most existing artificial intelligence for IT operations products on the market fall into two categories:
The first category is rule+data tool products. They use the AI label but are actually automation with preset rules. They fail immediately in border scenarios.
The other category uses the MCP/Workflow architecture. They have a Dependency on users to specify open source models. Their scenario coverage is limited. New scenarios are published slowly. They can hardly handle infrequent and complex problems.
DAS Agent achieves a qualitative breakthrough.
• Published 20+ O&M skills. AI can freely combine and invoke them. They are not limited by Workflow orchestration.
• Based on historical O&M ticket training, it is not a wrapper for a general Large Language Model (LLM). It shows a clear advantage in accuracy and stability.
• Diagnosis + operation closed loop: It directly completes O&M operations such as rate limiting and session management in the dialog box.
• 7×24 hour automatic O&M: It automatically scans managed instances under the account every day. It analyzes resource usage, slow SQL statements, deadlocks, space, and security risks one by one. This shifts the focus from firefighting to prevention.
• O&M daily report: It automatically runs diagnostics on instance health. It outputs a unified O&M report for multiple instances under the account. This supports cross-engine instances. The report includes high-value suggestions based on metric trends,comparative analysis, slow SQL statements, resource usage, and security risks.
• SQL optimization: It provides index recommendations, equivalent rewriting, and pre-publishing dry runs. The optimization effect is significantly better than traditional solutions. It improves query efficiency from the source.
• Intelligent Diagnosis: It covers scenarios such as abnormal CPU or memory, deadlocks, large keys, and lock contention. It accurately locates the root cause. It provides optimization suggestions and immediate remediation plans within seconds.
• Knowledge Q&A pair: It analyzes and compares cross-engine O&M reports. It covers multiple types such as abnormal handling suggestions, upgrade or downgrade suggestions, and change impact analysis. Expert-level knowledge Q&A pairs provide an instant response to change difficulties.
• Small and medium-sized teams: They have no full-time DBAs. Developers do O&M part-time. Or they spend a lot of money to hire external personnel for Support. DAS Agent can significantly reduce these costs. It is online 7×24 hours. It never asks for leave or resigns.
• Large enterprises: They have many instances and complex engines. DBAs are exhausted from firefighting for a long time. DAS Agent can significantly reduce their burden. It lets DBAs focus on truly valuable tasks.
Database O&M should no longer rely on simply adding manpower. DAS Agent proves that when AI gains realm depth, logical inference, and closed-loop operations, it truly evolves from a supporting tool into an autonomous entity.
DAS Agent now supports unified management for multicloud and multi-engine setups. It seamlessly integrates Alibaba Cloud cloud-native databases (RDS, PolarDB, Tair, and MongoDB) and self-managed or cross-cloud instances (MySQL, PG, Redis, and MongoDB) from any deployment. This gives enterprises a global database operations management hub that is highly compatible across any environment. It breaks data silos and enables one-stop O&M administration.
Database administrators (DBAs) no longer have to switch between multiple platforms and accounts. A single dialog box handles everything from diagnosis to operation in a complete closed loop.
The era of AI-native O&M has arrived!
Lindorm Tops VectorDBBench Performance, Defining New Heights for Vector Retrieval
Alibaba Cloud Indonesia - March 20, 2026
ApsaraDB - March 30, 2026
Alibaba Cloud Native - October 23, 2024
Alibaba Container Service - January 15, 2026
Alibaba Cloud New Products - June 1, 2020
Alibaba Cloud Native Community - November 6, 2025
Bastionhost
A unified, efficient, and secure platform that provides cloud-based O&M, access control, and operation audit.
Learn More
Managed Service for Grafana
Managed Service for Grafana displays a large amount of data in real time to provide an overview of business and O&M monitoring.
Learn More
Container Compute Service (ACS)
A cloud computing service that provides container compute resources that comply with the container specifications of Kubernetes
Learn More
Container Service for Kubernetes
Alibaba Cloud Container Service for Kubernetes is a fully managed cloud container management service that supports native Kubernetes and integrates with other Alibaba Cloud products.
Learn MoreMore Posts by ApsaraDB
