Cloud routing for supergraphs

When you create a cloud supergraph with GraphOS Studio, GraphOS automatically provisions and manages a router that acts as an entry point to your GraphQL APIs. In this architecture, individual GraphQL APIs are called subgraphs. Clients send operations to your router's public endpoint, instead of sending them directly to your subgraphs.

Initializing your router

Your router is provisioned automatically whenever you create a cloud supergraph in GraphOS Studio, or whenever you create a new variant for an existing cloud supergraph. Each variant has its own distinct router.

When you first create a variant, the router provisioning process can take a few minutes. If this process hasn't completed yet for a particular variant, an INITIATING ENDPOINT label is shown at the top of the variant's page in Studio:

Configuring your router

See Cloud router configuration.

Cloud launches

Publishing a new subgraph schema or editing a cloud router's configuration triggers a new launch. Every launch automatically deploys new router instances for your supergraph.

Router version updates

With cloud routing, GraphOS runs a fully managed fleet of Apollo Routers for you. Apollo manages the version of Apollo Router that is deployed and ensures that newly released versions are deployed to cloud routers within 30 days of release. Some minor and patch versions may be skipped.

Router releases go through rigorous testing before deployment on GraphOS Cloud. During the deployment of new versions, an Apollo engineer actively oversees the process and rolls back any cloud routers that fail to boot up. While some edge cases may arise—for example, a query planner update could result in slightly degraded performance—router updates should not disrupt your supergraphs.

Automatic deletion of unused routers

Apollo automatically deletes variants of Serverless cloud supergraphs that receive zero operations for 60 consecutive days. This deletes the router, along with all the variant's historical metrics.

• Apollo will notify you via email whenever a Serverless variant is approaching this 60-day limit.
• To prevent deletion, execute at least one GraphQL operation on the variant's router before the 60-day limit.

Out of memory errors

You may receive out of memory errors when using cloud routers on the Serverless plan. Since the Serverless plan offers limited underlying compute resources, Serverless cloud routing offers a fixed amount of throughput capacity.

Complex schemas may exceed available machine memory for Serverless cloud routers. For example, Serverless cloud routers don't support schemas with over 1,000 input types. Additionally, complex queries that return too much data or query too many subgraphs simultaneously may fail.

You can upgrade to Cloud Dedicated for control over scaling and performance. Get in touch for a free trial.

Cloud routing FAQ

What happens if my cloud router goes down?

If your supergraph's cloud router goes down, queries sent to it will fail until service is restored. However, every cloud router has redundant machines in place to help prevent this.

How is my data protected with cloud routing?

The entire GraphOS platform (including its cloud routing infrastructure) is SOC 2 Type 2 certified. Secrets are encrypted both in transit and at rest, and they are available only inside the runtime environment where users have total control over when those secrets are resolved in configuration.

The Apollo Router (the underlying technology for cloud routing) has been tested and audited by Doyensec, with no issues.

How are cloud routers hosted?

Apollo works with major public cloud providers to deliver GraphOS cloud routing.

Which regions are available for cloud routers?

Cloud routing for Serverless plans is only available in Chicago, USA. Our infrastructure has interconnects with major public cloud providers, so latency is minimal. For example, latency between Chicago and AWS in Virginia is approximately 10 ms. If there's a specific region you'd like us to support please request a region.

Cloud routing on the Dedicated plan has a wider variety of options. Get in touch to learn more.

Can I choose my cloud router's region?

Choice of region for cloud routers is only available on the Dedicated plan.

Does GraphOS also host my individual subgraphs?

No. GraphOS only hosts the runtime for your supergraph's cloud router. GraphQL servers for your individual subgraphs are still hosted in your infrastructure.

Which GraphQL server libraries are compatible with cloud routers?

Cloud supergraphs use Apollo Federation 2 for their core architecture. Many GraphQL server libraries support Federation 2.

Your GraphQL API does not already need to be using Apollo Federation to add it to a cloud supergraph.