A conceptual image showing a breadth-first data traversal, with multiple nodes being processed simultaneously in organized layers.

InfrastructureHigh

Shopify Rebuilds GraphQL For Speed

TL;DR: Shopify has launched GraphQL Cardinal, a new execution engine that uses a breadth-first approach instead of the traditional depth-first method. This change delivers significant performance gains, including up to 15x faster field execution, 6x lower garbage collection overhead, and improved latency for large-scale queries.

By Ashish KaleInfoQjust now1 min readupdated 3h ago

Source

Key facts

Category: Infrastructure
Impact: High
Published: just now
Source: InfoQ

Full summary

Shopify's new GraphQL engine uses a breadth-first approach to deliver up to 15x faster performance and significantly lower overhead.

Shopify has introduced a new GraphQL execution engine named Cardinal, completely redesigning its approach to processing queries. The new engine abandons the conventional depth-first traversal method in favor of a breadth-first execution model. This fundamental change is designed to improve performance for large-scale operations. The company reports significant gains, including field execution speeds up to 15 times faster, a six-fold reduction in garbage collection (GC) overhead, and a four-second improvement in P50 latency. Cardinal focuses on enhancing efficiency at the execution layer and is optimized for batched resolver processing, which is particularly effective for high-cardinality queries.

This architectural update is important because it demonstrates a novel solution to common performance bottlenecks in large GraphQL implementations. For companies relying on GraphQL, especially in sectors like e-commerce where queries can become complex, Shopify's results offer a compelling case study. The shift to a breadth-first model could inspire other organizations to re-evaluate their own GraphQL architectures. The reported improvements are highly relevant for developers, CTOs, and infrastructure teams seeking to build more scalable and efficient systems, highlighting a practical path to overcoming performance limitations.