The PolarDB-X AI assistant is an intelligent operations and maintenance tool for the cloud-native distributed database PolarDB-X. Powered by large language models and a distributed database expert knowledge base, it integrates three core capabilities: intelligent Q&A, intelligent diagnosis, and intelligent inspection. The AI assistant supports natural language interaction and has a deep understanding of the PolarDB-X distributed architecture, enabling it to perform root cause analysis for multi-node performance issues and provide recommendations for sharding optimization, SQL routing analysis, and elastic scaling. It uses RAM for access control to secure your instances and data. This significantly reduces the operational complexity and barrier to entry for distributed databases, helping you troubleshoot issues efficiently and manage large-scale clusters.
Core capabilities
Intelligent Q&AUse natural language to query data, topology, and performance metrics across multiple instances. This enables batch retrieval and issue diagnosis. The assistant also answers questions about resource assessment, billing, and best practices. | Intelligent diagnosisIntegrates the AI assistant into pages such as Slow SQL Statements, Deadlock Analysis, and Performance Trend. It combines distributed execution plans with contextual information to provide real-time root cause diagnosis and optimization recommendations. | Intelligent inspectionPerforms one-click health inspections of your instances, assessing their operational status, backup configurations, security settings, and high availability and disaster recovery capabilities to quickly identify potential risks and misconfigurations. |
Accessing the AI assistant
Log on to the PolarDB-X console. On the Instances page, click the
icon in the rightmost sidebar to open the chat interface.
Additional entry points: PolarDB-X also embeds the AI assistant into various feature pages:
Entry point
Example
Slow SQL Statements: On the page for the target instance, you can analyze slow SQL queries within a specified time range.
Deadlock Analysis: On the page for the target instance, you can analyze deadlocks within a specified time range.
Performance Trend: On the page for the target instance, you can analyze metrics such as CPU utilization, memory utilization, active sessions, or request response time within a specified time range.
To use the AI assistant, a RAM user must be granted the AliyunDRDSReadOnlyAccess (read-only permission for PolarDB-X) and AliyunYaoChiAgentAccess policies. For instructions on how to grant permissions, see Grant permissions to a RAM user.
AI assistant capability overview
The PolarDB-X AI assistant covers the following skill domains, providing comprehensive support for distributed database operations scenarios, from performance optimization to migration assessment:
Skill domain | Core capability | Use cases |
Fault diagnosis | Root cause identification and event analysis | Diagnose logs (error logs/slow queries) and troubleshoot connection and session exceptions to quickly identify root causes and assess their impact. |
Performance optimization | Instance performance diagnosis | Monitor CPU, memory, storage, and IOPS resource levels, analyze performance bottlenecks, and identify slow SQL queries with optimization recommendations. |
Capacity management | Storage and resource planning | Analyze storage space usage trends and growth forecasts to assess resource scaling needs and avoid capacity shortages. |
Risk governance | Version security and risk alerts |
|
Backup and recovery | Backup policy and disaster recovery assessment |
|
Security and compliance | Access control and audit | Verify IP address whitelist configurations, account permissions, SSL/TDE encryption status, and audit log integrity to ensure security and compliance. |
Asset management | Instance lifecycle management | Filter and query instances across multiple dimensions, receive subscription expiration reminders, and monitor the overall status of all instances for precise asset management. |
Instance inspection | Full-stack health check | Automatically inspect instance configurations, performance baselines, security policies, and backup status to generate inspection reports and provide remediation suggestions. |
Intelligent Q&A
The intelligent Q&A feature allows you to use natural language to quickly query instance information and ask product-related technical questions. The AI assistant can help you understand the distributed architecture of PolarDB-X and has expert knowledge of concepts such as compute nodes and data nodes.
Use cases
Global resource query: Perform batch queries for instance configurations, topologies, sharding rules, and performance metrics.
Anomalous instance screening: Identify at-risk instances with issues like abnormal connection counts on data nodes or compute nodes, shard hotspots, and storage skew.
Sharding and topology query: Query rules for database and table sharding, data distribution, and the status of any global secondary index.
Compliance and governance check: Query security policies, such as whether SQL audit is enabled or the status of IP address whitelist configurations and SSL encryption.
Product and technical consultation: Get information on PolarDB-X features, version details, and distributed best practices.
Quick start
By asking simple questions, the PolarDB-X AI assistant helps you quickly retrieve information about instance configurations, topologies, data distribution, and performance metrics.
Information you can query
Instance properties: Instance ID, version, specifications, status, number of compute and data nodes, storage type, and VPC and network information.
Topology: Lists and statuses of compute nodes, data nodes, and GMS nodes, along with read/write splitting configurations.
Data distribution: Rules for database and table sharding, sharding key information, data distribution, and the status of global secondary indexes.
Billing and cost: Billing method, expiration date, and the specifications and cost breakdown of each node.
Monitoring metrics: View metrics for each node in batches, such as CPU utilization, IOPS, QPS/TPS, and connection count.
Example questions
List all PolarDB-X instances under my account in the China (Hangzhou) region.
List the details of the compute and data nodes for instance
pxc-bp1abc****, including the specifications and running status of each node.Show the data distribution for the
order_dbdatabase in instancepxc-bp1abc****.Find all instances where the storage utilization of any data node exceeds 80%.
List all production PolarDB-X instances that do not have SQL audit enabled.
Intelligent diagnosis
Powered by large AI models, the AI assistant's intelligent diagnosis feature provides comprehensive, automated diagnosis and optimization capabilities for distributed databases. You can use the AI assistant to perform in-depth analysis of scenarios such as instance status, slow SQL queries, and storage space to quickly identify root causes and receive actionable optimization recommendations.
Use cases
Slow SQL diagnosis: Analyzes distributed query execution plans to automatically identify common performance issues like full table scans, cross-shard joins, and data skew. It also provides recommendations for index optimization and sharding key adjustments.
CPU, memory, and IO performance troubleshooting: When an instance experiences a CPU spike, abnormal memory usage, or an IO bottleneck, the AI assistant can analyze monitoring metrics and runtime logs to locate resource hotspots and provide targeted tuning solutions.
Quick start
Instance anomaly analysis: Supports analysis for scenarios such as CPU spikes, storage space growth, data node exceptions, distributed transaction timeouts, and slow queries.
Performance optimization: When CPU, memory, or IOPS metrics show abnormal fluctuations, it analyzes resource hotspots and provides targeted tuning solutions.
Capacity management: Analyzes storage space usage trends to provide growth forecasts and recommendations for scaling windows.
Intelligent inspection
Inspection coverage
The inspection report covers the following dimensions:
Instance status: Status of primary instances, read-only instances, and associated instances, as well as data node disk usage levels and maintenance windows.
Backup status: Backup policy configuration (type, frequency, retention period), backup set validity (latest backup time, success rate), and whether cross-region backup is enabled.
Security configuration: IP address whitelist access control, separation of duties for accounts, SSL encryption for data in transit, Transparent Data Encryption (TDE) for data at rest, and SQL audit status.
High availability and disaster recovery: Distribution of compute nodes across availability zones, multi-replica distribution of data nodes (Leader, Follower, Logger), and an overall assessment of cross-availability zone disaster recovery capabilities.
Quick start
Log on to the PolarDB-X console. On the Instances page, select the target instance and click AI Instance Inspection in the upper-right corner of the instance details page.
In the AI assistant interface, you can configure the inspection items and time range based on your business requirements.
After the task is complete, you can view the detailed inspection results and optimization recommendations for the target instance. You can also click the Download button in the lower-right corner to export the report as a PDF or image file.
Billing
The PolarDB-X AI assistant is currently in a free public beta. You can use all its features at no charge.
Usage notes
Be specific in your questions: When asking questions, provide as much detail as possible, including specific instance IDs, the instance's region, a clear time range, and a precise description of the issue. For distributed scenarios, specify the relevant data node or database name for a more accurate diagnosis.
Use conversational context: The AI assistant supports multi-turn conversations and can follow up on previous diagnostic results. However, always carefully evaluate suggestions that involve sensitive operations, such as changing sharding rules, before taking action.
Timeliness of diagnosis: Performance diagnosis relies on monitoring data. Initiate the analysis as soon as possible after an issue occurs to ensure that the relevant metric data is still within its retention period. If the session becomes unresponsive, exit and start a new conversation.
Review recommendations carefully: Optimization recommendations are generated based on current monitoring data and model inference and are for reference only. Before applying any critical changes, especially those affecting the distributed architecture, thoroughly test and evaluate them.
Security and compliance
Data access scope: To provide accurate diagnoses and answers, the PolarDB-X AI assistant accesses critical information such as instance metadata, topology, performance monitoring metrics, and the slow query log, as well as the questions you enter in the chat. However, it does not execute any DDL or DML statements, directly modify instance configurations, or trigger data migration.
Permission control: The AI assistant strictly inherits the RAM permission model. It directly performs read-only operations, but any recommendations involving changes require your manual confirmation before execution.
Data privacy: Your query history and cluster data are handled in strict compliance with Alibaba Cloud's privacy policy. This data is not used for model training or shared with any third parties.
Disclaimer: All diagnostic conclusions and optimization suggestions, especially those involving SQL statements and parameter modifications, are provided for reference only. Before you apply any changes in a production environment, especially distributed operations like sharding rule adjustments or global secondary index modifications, a professional database administrator (DBA) or technical expert must thoroughly review and assess them.