Trino Configuration

The Query Engine connects to Trino through a configurable connection layer that manages JDBC connections, session properties, resource groups, and query timeouts. Configuration is managed through environment variables and application properties.

Connection Settings

Property	Default	Description
`trino.host`	`trino-coordinator`	Trino coordinator hostname
`trino.port`	`8080`	Trino coordinator port
`trino.catalog`	`delta`	Default catalog for queries
`trino.schema`	`default`	Default schema for queries
`trino.source`	`matih-query-engine`	Source identifier for query tracking

Connection Pool

The JDBC connection pool is configured for high-concurrency query workloads:

Property	Default	Description
`trino.pool.maxConnections`	50	Maximum connections in the pool
`trino.pool.minIdle`	10	Minimum idle connections
`trino.pool.connectionTimeout`	30000	Connection timeout in milliseconds
`trino.pool.idleTimeout`	600000	Idle connection timeout

Session Properties

Per-query session properties control Trino's execution behavior:

sessionProperties.put("query_max_run_time", "5m");
sessionProperties.put("query_max_memory", "1GB");
sessionProperties.put("join_distribution_type", "AUTOMATIC");
sessionProperties.put("enable_dynamic_filtering", "true");

Tenant-specific session properties are applied based on the tenant's tier:

Tier	Max Run Time	Max Memory	Max Concurrent
FREE	2 min	256MB	2
STARTER	5 min	512MB	5
PROFESSIONAL	15 min	2GB	20
ENTERPRISE	60 min	8GB	50

Resource Groups

Trino resource groups enforce concurrency and memory limits per tenant:

{
  "rootGroups": [
    {
      "name": "tenant_${TENANT_ID}",
      "softMemoryLimit": "20%",
      "maxQueued": 100,
      "softConcurrencyLimit": 10,
      "hardConcurrencyLimit": 20,
      "schedulingPolicy": "fair"
    }
  ]
}

Health Checks

The Query Engine monitors Trino availability through periodic health checks against the coordinator's /v1/info endpoint. If Trino becomes unavailable, queries are queued until connectivity is restored.

Overview Catalogs