Struct GCRealtimeDatabase 

pub struct GCRealtimeDatabase {
    datapoint_db: RwLock<Vec<Mutex<GCDatapointState>>>,
    subscriber_identifier: GCSubscriberIdentifier,
}

The RealtimeDatabase that keeps track of the last datapoint value updated. This database is designed for high frequency writes and low reads.

It uses a Mutex, from the [parking_lot] crate, per each datapoint. Following the study, a spin lock is slower most of the times, when compared to a Mutex due to this reason it was decided to use it. Additionally, a lock-free mechanism could in theory be used, but this can cause a big implementation/maintenance effort in the future, in case there is a need to add new metrics or keep small buffer of values. A simple Mutex for each datapoint should keep the code simple and efficient as contention is expected to be low.

Usage

let db = GCRealtimeDatabase::new();
db.add_datapoint(0,1);

let dpv = gc_alloc_datapoint(GCDatapointValue::new_u64(0, 10, 1725271559533000000, 0.into()));
db.update_datapoint_value(dpv.clone()).unwrap();

assert_eq!(db.get_datapoint_state(0).unwrap().get_latest_value().deref(), dpv.deref());

Fields

datapoint_db: RwLock<Vec<Mutex<GCDatapointState>>>

subscriber_identifier: GCSubscriberIdentifier

Implementations

impl GCRealtimeDatabase

pub fn new() -> Self

pub fn capacity(&self) -> usize

Returns the number of datapoints that can fit in the database without resizing.

pub fn new_empty(datapoint_count: usize) -> Self

Creates a new empty RealtimeDatabase with a given number of datapoints. The id of the datapoints will be the index of the datapoint in the database.

pub fn clear(&self)

Clears the database.

pub fn clear_datapoint( &self, datapoint_id: GCDatapointID, ) -> Result<(), GCRealtimeDatabaseError>

Uninitialize a datapoint. This clears every past values stored in the datapoint and sets it to uninitialized.

pub fn add_datapoint(&self, datapoint_id: GCDatapointID, capacity: u64)

If the datapoint_id is greater than the current size of the database, it will resize the database to the new size.

pub fn update_datapoint_value( &self, value: GCArc<GCDatapointValue>, ) -> Result<(), GCRealtimeDatabaseError>

Sets the value of a datapoint. The value is only updated if the timestamp is greater than the current value already stored.

pub fn get_datapoint_state( &self, datapoint_id: u64, ) -> Result<GCDatapointState, GCRealtimeDatabaseError>

Returns the state of a datapoint.

pub fn for_each_state<F>(&self, f: F)

where F: FnMut(GCDatapointID, &GCDatapointState),

Returns an iterator over all datapoint states. Only initialized datapoints are iterated.

pub fn get_datapoint_state_ref( &self, datapoint_id: u64, ) -> Result<GCDatapointStateRefLock<'_>, GCRealtimeDatabaseError>

Returns a reference to the Datapoint state, unlike GCRealtimeDatabase::get_datapoint_state this method does not clone the state and is more efficient. Although, the returned reference is behind a mutex that only get’s released when the reference goes out of scope.

Trait Implementations

impl Default for GCRealtimeDatabase

fn default() -> Self

Returns the “default value” for a type.

impl GCDatapointSubscriber for GCRealtimeDatabase

fn update_datapoint_value( &self, data_value: GCArc<GCDatapointValue>, ) -> Result<(), GCMessageBrokerError>

Sends a datapoint value to the Plugin, this calls issues a call to gc_plugin_receive_datapoint exported by the plugin

fn unique_identifier(&self) -> &GCSubscriberIdentifier

Returns the unique identifier of the subscriber