The Mini DBA AI Assistant helps database teams interpret live monitoring data, performance symptoms, execution plans, alerts, and estate-level health from inside the Mini DBA Console. It is designed for practical database troubleshooting: ask a question from the page you are already viewing, choose which context to include, and get an explanation or next-step plan that uses the monitoring evidence Mini DBA has already collected.
What The AI Assistant Does
The AI Assistant is available from many server detail pages through the Ask AI panel. When you open it from a SQL Server, Azure SQL, PostgreSQL, MySQL, MariaDB, or Oracle page, Mini DBA can build a live context snapshot for that server and pass it with your question.
The assistant can help with:
- Explaining performance symptoms from the current page.
- Summarizing resource pressure, waits, sessions, and alerts.
- Suggesting investigation steps for active incidents.
- Reviewing query and execution plan evidence where plan context is available.
- Comparing current values with the page and time range you selected.
- Producing a DBA-friendly action plan for the next person handling the issue.
The AI Assistant does not replace database expertise or change the database for you. Treat the response as guided analysis. Confirm important recommendations using Mini DBA evidence, change control, and your normal database administration tools.
How Page Context Is Passed To AI
When you ask a question from a Mini DBA detail page, the assistant can include a snapshot that identifies the server, database platform, connection status, current page, and time context. The prompt tells the AI provider that the snapshot is live Mini DBA data already collected by the application, so the model should analyze the supplied evidence rather than saying it cannot access live data.
The context can include:
- Server name and database platform.
- Whether the monitored server is connected.
- Last refresh time or selected historical time range.
- The page you were viewing, such as Activity, Waits, Memory, I/O, Query Store, Indexes, Deadlocks, or Health Checks.
- Optional metrics, sessions, and alerts depending on the switches in the AI panel.
- Additional detail-page context, such as an execution plan, SQL text, elastic pool context, or other page-specific data passed by the console.
If a live snapshot is not available yet, Mini DBA shows a message asking you to wait for the next refresh and try again. This usually means the page has loaded before the server model or current server-list snapshot has enough data.
Context Options
The Ask AI panel includes context options so you can control how much evidence is sent.
| Option | What It Adds | When To Use It |
|---|---|---|
| Metrics | CPU, memory, waits, I/O, service-tier, host, database, and model metrics where available. | Use for resource pressure, baseline comparison, and general performance analysis. |
| Sessions | Active sessions, requests, blockers, blocked sessions, MySQL or PostgreSQL session lists, and similar live workload rows where available. | Use for blocking, long running queries, session pile-ups, and query-level triage. |
| Alerts | Active alerts, severity, detected values, thresholds, and alert descriptions. | Use when you want the AI response to explain why Mini DBA is alerting and what to do next. |
Metrics and sessions are enabled by default in the page-level assistant. Alerts are optional by default because alert context can add volume and may include sensitive operational messages. Enable Alerts when the active alert list is relevant to the question.
Detail Page And Execution Plan Context
Several Mini DBA detail pages pass extra context to the AI Assistant. This is especially useful for query and plan analysis because the general server snapshot alone may not contain the exact SQL text, plan shape, expensive operators, or selected row. For the full plan workflow, see Execution Plan Analysis.
Detail-page context may include:
- SQL Server execution plan XML from Activity, Query Store, cached SQL, or database query pages.
- Azure SQL execution plan and Query Store context.
- PostgreSQL plan text or XML where available.
- MySQL execution plan tree, JSON, or operator list where available.
- Oracle SQL text, plan text, or parsed plan operators where available.
- Azure elastic pool summary context when using the elastic pool page.
When additional context is present, Mini DBA appends it under an Additional Context section before sending the request. Ask focused questions such as "which operator should I investigate first?", "what evidence suggests missing indexes?", or "what could explain the wait pattern in this plan?".
Default AI And BYO AI
Mini DBA can use the default AI provider where configured. The default provider includes a limited daily free token allowance per user. The AI panel shows daily token usage when the default provider is active.
Bring Your Own AI, or BYO AI, lets you use your own AI subscription and model from AI Provider Settings. BYO AI is not limited by the default free daily quota because requests are sent to your configured provider. Supported provider choices in the console include OpenAI, Azure OpenAI, OpenAI-compatible APIs, Ollama, and Anthropic Claude.
For setup steps, see Configure BYO AI.
Free Daily Token Limit
The default free daily token limit is designed to make AI available without requiring every user to configure a provider first. The limit is fixed in the product at 100,000 tokens per user per day for the default provider.
Token use is tracked in the browser for the default provider. If the free daily threshold is reached, Mini DBA explains that the allowance resets tomorrow and links to BYO AI settings. BYO AI uses your own subscription and does not consume the default free token allowance.
For token, privacy, and data-sharing guidance, see AI Context, Privacy, And Tokens.
Popular AI Workflows
Use these workflow pages when you want a guided example:
- Configure BYO AI to connect OpenAI, Azure OpenAI, Ollama, Anthropic, or an OpenAI-compatible provider.
- Performance Tuning With AI to use Mini DBA metrics, sessions, waits, Query Store, and execution plans.
- Alert Triage With AI to turn active alerts into an investigation plan.
- AI Context, Privacy, And Tokens to understand what is sent, how quota works, and how to reduce sensitive context.
Best Practices
- Ask specific questions tied to the page you are viewing.
- Include Metrics for resource questions and Sessions for live workload questions.
- Include Alerts when the alert list is part of the incident.
- Use execution plan AI from a plan detail page when the question is query-specific.
- Keep prompts short and ask follow-up questions in the same panel to use conversation history.
- Do not paste secrets, passwords, connection strings, customer credentials, or private keys into the prompt.
- Use BYO AI when your organization requires a specific AI provider, private endpoint, regional policy, or model.
- Treat recommendations as evidence-led suggestions, not automatic change approval.
Related Pages
- Configure BYO AI
- Performance Tuning With AI
- Alert Triage With AI
- AI Context, Privacy, And Tokens
- Execution Plan Analysis
- AI Provider Settings
- MCP Server Settings
- SQL Server Performance Tuning
- Alerts Page