WARNING

How to Fix Azure SQL ServerBusy (DTU Limit Reached)

Quick Fix Summary

TL;DR

Immediately scale up your Azure SQL Database tier or service objective to increase DTU capacity.

The 'ServerBusy' error occurs when your Azure SQL Database exhausts its DTU (Database Transaction Unit) quota, triggering resource governance and throttling. This is a performance-based throttle, not a complete outage, but it will cause application timeouts and failures.

Diagnosis & Causes

Sudden, unanticipated spike in query workload or user traffic.

Inefficient queries (missing indexes, table scans) consuming excessive resources.

Long-running transactions or blocking locks monopolizing DTUs.

Database tier (e.g., Basic, Standard) is undersized for the production load.

Concurrent processes like ETL jobs or reporting queries running during peak hours.

Recovery Steps

Step 1: Immediate Mitigation - Scale Up via Azure Portal

Quickly provision more DTUs to relieve pressure. This is the fastest way to restore service while you diagnose the root cause.

bash

# 1. Navigate to your SQL Database in the Azure Portal.
# 2. Click 'Scale' under 'Settings'.
# 3. Select a higher service tier (e.g., from Standard S3 to Premium P2).
# 4. Click 'Apply'. The scale operation typically completes in seconds to minutes.

Step 2: Identify the Resource-Intensive Query

Use Azure SQL's built-in Query Performance Insight or DMVs to pinpoint the query consuming the most DTUs.

sql

-- Find top 5 queries by average CPU consumption (a primary DTU component)
SELECT TOP 5
    qs.query_hash,
    SUM(qs.avg_cpu_time * qs.execution_count) / 1000.0 AS total_cpu_sec,
    SUM(qs.avg_logical_io_reads * qs.execution_count) AS total_logical_reads,
    qt.query_sql_text
FROM sys.query_store_query_text qt
JOIN sys.query_store_query q ON qt.query_text_id = q.query_text_id
JOIN sys.query_store_plan p ON q.query_id = p.query_id
JOIN sys.query_store_runtime_stats qs ON p.plan_id = qs.plan_id
WHERE qs.last_execution_time > DATEADD(hour, -1, GETUTCDATE())
GROUP BY qs.query_hash, qt.query_sql_text
ORDER BY total_cpu_sec DESC;

Step 3: Implement Short-Term Query Optimization

Apply immediate fixes to the problematic query identified in Step 2, such as adding missing indexes or rewriting.

sql

-- Example: Create a covering index for a high-cost query performing a scan.
-- Analyze the query's WHERE clause and JOIN columns.
CREATE NONCLUSTERED INDEX IX_YourTable_Covering
ON dbo.YourTable (JoinColumn, FilterColumn)
INCLUDE (SelectedColumn1, SelectedColumn2);
-- Then, clear the plan cache to force recompilation with the new index.
ALTER DATABASE SCOPED CONFIGURATION CLEAR PROCEDURE_CACHE;

Step 4: Configure Proactive Alerts and Auto-Scale

Prevent future outages by setting up alerts for high DTU and configuring auto-scale rules where appropriate.

bash

# Create an alert for DTU percentage > 80% using Azure CLI
az monitor metrics alert create -n 'HighDTUAlert' \
--resource-group YourResourceGroup \
--scopes /subscriptions/YourSub/resourceGroups/YourRG/providers/Microsoft.Sql/servers/YourServer/databases/YourDB \
--condition "avg percentage_dtu_used > 80" \
--description "Alert for high DTU consumption" \
--action /subscriptions/YourSub/resourceGroups/YourRG/providers/microsoft.insights/actionGroups/YourActionGroup

Architect's Pro Tip

"For persistent, spiky loads, switch to the vCore model and enable Hyperscale. It decouples compute and storage, allowing near-instantaneous independent scaling and read-scale replicas to offload reporting queries."

Frequently Asked Questions

What's the difference between 'ServerBusy' and a complete connection failure?

'ServerBusy' is a throttling response; some connections/queries may still succeed intermittently. A complete failure (like 'database does not exist') indicates a more severe configuration or network issue.

How long does it take for a scale-up operation to take effect?

Scaling within the same tier (e.g., S1 to S3) is typically online and completes in seconds. Changing tiers (e.g., Standard to Premium) may take a few minutes and involves a brief reconnection drop.

Can I automatically scale my Azure SQL Database?

Yes, using Azure Automation runbooks or Logic Apps triggered by metrics alerts. However, for predictable daily patterns, consider using the 'Scheduled Scaling' feature in the Hyperscale service tier.

Related Azure Guides

SqlTimeoutException