Technologies/Redis/client_backend_usage
RedisRedisMetric

client_backend_usage

The number of active connections to the SQL Warehouse.
Dimensions:None
Available on:OpenTelemetryOpenTelemetry (1)

Summary

Represents the current utilization of the client backend connection pool, typically as a count or percentage. High usage indicates active connection consumption, while consistent high values suggest potential pool exhaustion. This metric is critical for identifying connection pool starvation patterns that can block application threads waiting for available connections.

Interface Metrics (1)
OpenTelemetryOpenTelemetry
Number of connections currently in the connection pool state (used, idle)
Dimensions:None
Knowledge Base (1 documents, 0 chunks)
best practicesManage database connections  |  Cloud SQL for SQL Server  |  Google Cloud Documentation2265 wordsscore: 0.65This page provides best practices and code samples for managing database connections to Cloud SQL for SQL Server. It covers connection pooling, proper opening/closing of connections, connection count management, and exponential backoff strategies with examples in multiple programming languages.

Technical Annotations (3)

Configuration Parameters (1)
max_connections
Adjust dynamically based on workload to prevent queuing
Technical References (2)
PgBouncercomponentProxySQLcomponent
Related Insights (5)
Bolt Connection Pool Saturationwarning

High bolt connections combined with client backend wait time indicates connection pool exhaustion, causing clients to queue for available connections and degrading application response times.

Client-Side Connection Pool Misconfigurationwarning

High client_backend_wait_time combined with low client_backend_usage indicates application connection pool misconfiguration or excessive connection churn, even when database-side cloudsql_googleapis_com/database/network/connections appears normal.

Connection Queue Buildup Indicates Connection Stormwarning

When client_backend_create_time and client_backend_wait_time are elevated, new connections are queuing because the instance is saturated with connection requests. This often precedes connection pool exhaustion and instance instability.

Databricks Connect requires local client performance monitoringinfo
Connection pool exhaustion causing query queuing and lag spikeswarning