API7 Solution: High Availability for B2B Services

In the course of client communication, a key question frequently arises: "Do you provide high availability? and how? "

API7 Enterprise provides building blocks for customer-managed high-availability deployments. Availability depends on the customer's deployment architecture, infrastructure dependencies, and operations; self-hosted API7 Enterprise does not carry a product-level uptime percentage.

In the business environment, the high availability of API services is particularly critical as it directly impacts client continuity and reliability. Why is high availability so crucial for B2B business? Because, as a key metric, any interruption or failure of API services during critical moments can severely impact client business, resulting not only in financial losses but also potentially damaging client reputation and credibility.

How Does API7 Support Highly Available Deployments?

High availability is both a technical and business objective. The following sections describe API7 Enterprise building blocks that can support a resilient deployment; these mechanisms do not by themselves establish a service-availability commitment.

Stateless Control Plane

The API7 Enterprise control plane uses stateless application components, allowing instances to be restarted or scaled independently. This design can support fault recovery and resource scheduling, while the availability achieved still depends on the database, load balancing, infrastructure, and customer operations.

PostgreSQL as Default Configuration Center

API7 Enterprise uses PostgreSQL as its default configuration store rather than the etcd used by APISIX. PostgreSQL offers established replication and failover patterns, but customers must select, configure, monitor, and test an appropriate topology. A standby database can reduce recovery time after a primary-node failure; it does not by itself guarantee configuration or service availability.

Stateless Data Plane

The data plane processes business traffic and is based on APISIX. Its stateless components can be scaled or replaced. When combined with redundant instances, health checks, sufficient capacity, and load balancing, this design can reduce the impact of an instance failure; service continuity still depends on the complete customer-managed deployment.

Independent Data Plane and Control Plane

Separating the data plane from the control plane reduces the request path’s dependency on control-plane availability. A running data-plane instance keeps its configuration in memory and can continue using that last available configuration during some control-plane disruptions. New configuration changes, new instances, and other infrastructure failures can still affect service.

Usage Scenarios and Benefits of High-Availability Architecture

Customers can deploy redundant instances behind a load balancer with health checks, or use Kubernetes health checks and replacement behavior. These mechanisms can detect and replace some failed components, but their effectiveness depends on correct configuration, spare capacity, infrastructure dependencies, and operational testing.

When these mechanisms are appropriately designed, operated, and tested, they can reduce the impact of some failures, speed recovery, and help continue processing part of the traffic. They do not prevent every business interruption or guarantee lower maintenance costs. Outcomes depend on recovery objectives, redundancy, external dependencies, monitoring, and customer operating procedures.

Conclusion

In summary, API7 Enterprise provides high-availability mechanisms such as control-plane and data-plane separation, stateless component scaling, and flexible deployment models. When appropriately designed and operated, these mechanisms can reduce failure impact and support recovery during expected failure scenarios.

These mechanisms do not guarantee uninterrupted service. Achieved availability depends on the customer's deployment architecture, component redundancy, infrastructure dependencies, monitoring, recovery procedures, and operations.