which is different from Envoy. Envoy takes the overprovisioning into
account in both [locality-weighted load balancing](https://www.envoyproxy.io/docs/envoy/latest/intro/arch_overview/upstream/load_balancing/locality_weight)
and [priority failover](https://www.envoyproxy.io/docs/envoy/latest/intro/arch_overview/upstream/load_balancing/priority),
but gRPC assumes that the xDS server will update it to redirect traffic
when this kind of graceful failover is needed. gRPC will send the
An [`lb_endpoint`](https://github.com/envoyproxy/envoy/blob/12a4bc430eaf440ceb0d11286cfbd4c16b79cdd1/api/envoy/api/v2/endpoint/endpoint_components.proto#L72)
Request matching based on:<ul><li>[Path](https://www.envoyproxy.io/docs/envoy/latest/api-v2/api/v2/route/route_components.proto#route-routematch) (prefix, full path and safe regex)</li><li>[Headers](https://www.envoyproxy.io/docs/envoy/latest/api-v2/api/v2/route/route_components.proto#route-headermatcher)</li></ul>Request routing to multiple clusters based on [weights](https://www.envoyproxy.io/docs/envoy/latest/api-v2/api/v2/route/route_components.proto#route-weightedcluster) | [A28](https://github.com/grpc/proposal/blob/master/A28-xds-traffic-splitting-and-routing.md) | v1.31.0 | v1.31.0 | v1.31.0 |