Streaming Query Results

Documentation

• Snowflake SDK Documentation

Streaming allows you to process large datasets without loading all results into memory at once, preventing out-of-memory errors and improving efficiency.

Key requirement: You must set streamResult: true in the execute options to enable streaming.

Modern Streaming with for await

Traditional streaming uses event handlers like on("data") and on("end"). This example demonstrates the much simpler for await pattern, which makes streaming as easy as iterating over an array.

Prerequisites

• You’ve established a connection to Snowflake and promisified it.

Example

// Optional: TypeScript interface for strongly-typed results
interface Customer {
  C_CUSTKEY: number;
  C_NAME: string;
  C_PHONE: string;
  C_ACCTBAL: number;
}

// Execute query with streamResult: true (this enables streaming)
const { statement } = connection.execute<Customer>({
  sqlText: `
    SELECT C_CUSTKEY, C_NAME, C_PHONE, C_ACCTBAL
    FROM SNOWFLAKE_SAMPLE_DATA.TPCH_SF1.CUSTOMER
    WHERE C_MKTSEGMENT = ?
  `,
  binds: ['AUTOMOBILE'],
  streamResult: true
});

// Create a stream (optionally specify start/end for a subset)
const stream = statement.streamRows({
  start: 250,
  end: 300
});

// Process results one row at a time with `for await`
for await (const row of stream) {
  console.log(`Customer: ${row.C_NAME}, Balance: ${row.C_ACCTBAL}`);
}