Struct scylla::transport::load_balancing::DefaultPolicy

pub struct DefaultPolicy { /* private fields */ }

The default load balancing policy.

It can be configured to be datacenter-aware, rack-aware and token-aware. Datacenter failover (sending query to a node from a remote datacenter) for queries with non local consistency mode is also supported.

Latency awareness is available, although not recommended: the penalisation mechanism it involves may interact badly with other mechanisms, such as LWT optimisation. Also, the very tactics of penalising nodes for recently measures latencies is believed to not be very stable and beneficial. The number of in-flight requests, for instance, seems to be a better metric showing how (over)loaded a target node/shard is. For now, however, we don’t have an implementation of the in-flight-requests-aware policy.

Implementations

impl DefaultPolicy

pub fn builder() -> DefaultPolicyBuilder

Creates a builder used to customise configuration of a new DefaultPolicy.

Trait Implementations

impl Debug for DefaultPolicy

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Default for DefaultPolicy

fn default() -> Self

Returns the “default value” for a type.

impl LoadBalancingPolicy for DefaultPolicy

fn pick<'a>( &'a self, query: &'a RoutingInfo<'_>, cluster: &'a ClusterData, ) -> Option<(NodeRef<'a>, Option<Shard>)>

Returns the first node to contact for a given query.

fn fallback<'a>( &'a self, query: &'a RoutingInfo<'_>, cluster: &'a ClusterData, ) -> FallbackPlan<'a>

Returns all contact-appropriate nodes for a given query.

fn name(&self) -> String

Returns the name of load balancing policy.

fn on_query_success( &self, _routing_info: &RoutingInfo<'_>, latency: Duration, node: NodeRef<'_>, )

Invoked each time a query succeeds.

fn on_query_failure( &self, _routing_info: &RoutingInfo<'_>, latency: Duration, node: NodeRef<'_>, error: &QueryError, )

Invoked each time a query fails.