Compute Engine v1 API - Enum BackendService.Types.LocalityLbPolicy (3.1.0)

public enum BackendService.Types.LocalityLbPolicy

Reference documentation and code samples for the Compute Engine v1 API enum BackendService.Types.LocalityLbPolicy.

The load balancing algorithm used within the scope of the locality. The possible values are: - ROUND_ROBIN: This is a simple policy in which each healthy backend is selected in round robin order. This is the default. - LEAST_REQUEST: An O(1) algorithm which selects two random healthy hosts and picks the host which has fewer active requests. - RING_HASH: The ring/modulo hash load balancer implements consistent hashing to backends. The algorithm has the property that the addition/removal of a host from a set of N hosts only affects 1/N of the requests. - RANDOM: The load balancer selects a random healthy host. - ORIGINAL_DESTINATION: Backend host is selected based on the client connection metadata, i.e., connections are opened to the same address as the destination address of the incoming connection before the connection was redirected to the load balancer. - MAGLEV: used as a drop in replacement for the ring hash load balancer. Maglev is not as stable as ring hash but has faster table lookup build times and host selection times. For more information about Maglev, see https://ai.google/research/pubs/pub44824 This field is applicable to either: - A regional backend service with the service_protocol set to HTTP, HTTPS, or HTTP2, and load_balancing_scheme set to INTERNAL_MANAGED. - A global backend service with the load_balancing_scheme set to INTERNAL_SELF_MANAGED, INTERNAL_MANAGED, or EXTERNAL_MANAGED. If sessionAffinity is not configured—that is, if session affinity remains at the default value of NONE—then the default value for localityLbPolicy is ROUND_ROBIN. If session affinity is set to a value other than NONE, then the default value for localityLbPolicy is MAGLEV. Only ROUND_ROBIN and RING_HASH are supported when the backend service is referenced by a URL map that is bound to target gRPC proxy that has validateForProxyless field set to true.

Namespace

Google.Cloud.Compute.V1

Assembly

Google.Cloud.Compute.V1.dll

Fields

Name Description
InvalidLbPolicy
LeastRequest

An O(1) algorithm which selects two random healthy hosts and picks the host which has fewer active requests.

Maglev

This algorithm implements consistent hashing to backends. Maglev can be used as a drop in replacement for the ring hash load balancer. Maglev is not as stable as ring hash but has faster table lookup build times and host selection times. For more information about Maglev, see https://ai.google/research/pubs/pub44824

OriginalDestination

Backend host is selected based on the client connection metadata, i.e., connections are opened to the same address as the destination address of the incoming connection before the connection was redirected to the load balancer.

Random

The load balancer selects a random healthy host.

RingHash

The ring/modulo hash load balancer implements consistent hashing to backends. The algorithm has the property that the addition/removal of a host from a set of N hosts only affects 1/N of the requests.

RoundRobin

This is a simple policy in which each healthy backend is selected in round robin order. This is the default.

UndefinedLocalityLbPolicy

A value indicating that the enum field is not set.

WeightedMaglev

Per-instance weighted Load Balancing via health check reported weights. If set, the Backend Service must configure a non legacy HTTP-based Health Check, and health check replies are expected to contain non-standard HTTP response header field X-Load-Balancing-Endpoint-Weight to specify the per-instance weights. If set, Load Balancing is weighted based on the per-instance weights reported in the last processed health check replies, as long as every instance either reported a valid weight or had UNAVAILABLE_WEIGHT. Otherwise, Load Balancing remains equal-weight. This option is only supported in Network Load Balancing.