Skip to main content

Resilience and Autoscaling

FoundryDB Edge is designed to keep your application reachable through node failures, whole-location outages, and shifting traffic. This page describes the behavior you can observe as a customer. Resilience and autoscaling are handled by the platform, so there is nothing to configure for them to take effect.

Request Path​

Each point of presence (PoP) serves your domain from a highly available pair: an active node that handles live traffic, plus a hot standby that is ready to take over. Requests reach the nearest point of presence, and the gateway forwards them to your application origin.

The serving address is what your domain resolves to. It stays stable across the events described below, which means your domain keeps pointing at the same address and your TLS certificate keeps working.

Node Failover Within a Location​

If the active node in a point of presence fails, the serving address moves to the hot standby within seconds. Your domain keeps resolving to the same address, so:

  • There is no DNS change.
  • There is no certificate re-issue.
  • Clients continue connecting to the same address.

After failover, the platform automatically works to restore the pair back to full redundancy, so the location returns to having both an active node and a hot standby.

Failover and recovery are automatic, but they are not instantaneous. Design your application and clients to tolerate brief connection interruptions and to retry transient errors.

Whole-Location Resilience​

If an entire point of presence becomes unavailable, your application's traffic is automatically repointed to another point of presence so your app stays reachable. When the original location recovers, traffic returns to its home location.

Because traffic can be served from a different European location during an outage, requests may travel a little further during that window. This is expected and resolves once the home location is healthy again.

Autoscaling​

Each point of presence automatically scales out when request volume rises and scales back in when it subsides, within configured bounds. You do not need to provision capacity ahead of a traffic spike or scale down manually afterward.

Scaling responds to sustained changes in traffic rather than momentary fluctuations, and it operates within the bounds configured for your edge service. Scaling is automatic, but it is not unbounded or instantaneous.

Self-Healing​

The platform continuously checks each point of presence against its desired capacity and reconciles any difference. If a node is lost transiently, the location recovers on its own without manual intervention.

What This Means For You​

  • Your domain resolves to a stable serving address that does not change during node failover.
  • Failover happens within seconds, with no DNS change and no certificate re-issue.
  • A whole-location outage does not take your app offline, because traffic is repointed to another location.
  • Capacity follows your traffic automatically, within configured bounds.
  • Transient node loss recovers on its own.

For the full feature set of the gateway, see the Edge Overview.