Crate ckb_metrics_runtime

High-speed metrics collection library.

metrics-runtime provides a generalized metrics collection library targeted at users who want to log metrics at high volume and high speed.

Design

The library follows a pattern of "senders" and a "receiver."

Callers create a Receiver, which acts as a registry for all metrics that flow through it. It allows creating new sinks as well as controllers, both necessary to push in and pull out metrics from the system. It also manages background resources necessary for the registry to operate.

Once a Receiver is created, callers can either create a Sink for sending metrics, or a Controller for getting metrics out.

A Sink can be cheaply cloned, and offers convenience methods for getting the current time as well as getting direct handles to a given metric. This allows users to either work with the fuller API exposed by Sink or to take a compositional approach and embed fields that represent each particular metric to be sent.

A Controller provides both a synchronous and asynchronous snapshotting interface, which is metrics-core compatible for exporting. This allows flexibility in integration amongst traditional single-threaded or hand-rolled multi-threaded applications and the emerging asynchronous Rust ecosystem.

Performance

Users can expect to be able to send tens of millions of samples per second, with ingest latencies at roughly 65-70ns at p50, and 250ns at p99. Depending on the workload -- counters vs histograms -- latencies may be even lower, as counters and gauges are markedly faster to update than histograms. Concurrent updates of the same metric will also cause natural contention and lower the throughput/increase the latency of ingestion.

Metrics

Counters, gauges, and histograms are supported, and follow the definitions outlined in metrics-core.

Here's a simple example of creating a receiver and working with a sink:

use metrics_runtime::Receiver;
use std::{thread, time::Duration};
let receiver = Receiver::builder().build().expect("failed to create receiver");
let mut sink = receiver.sink();

// We can update a counter.  Counters are monotonic, unsigned integers that start at 0 and
// increase over time.
sink.increment_counter("widgets", 5);

// We can update a gauge.  Gauges are signed, and hold on to the last value they were updated
// to, so you need to track the overall value on your own.
sink.update_gauge("red_balloons", 99);

// We can update a timing histogram.  For timing, we're using the built-in `Sink::now` method
// which utilizes a high-speed internal clock.  This method returns the time in nanoseconds, so
// we get great resolution, but giving the time in nanoseconds isn't required!  If you want to
// send it in another unit, that's fine, but just pay attention to that fact when viewing and
// using those metrics once exported.  We also support passing `Instant` values -- both `start`
// and `end` need to be the same type, though! -- and we'll take the nanosecond output of that.
let start = sink.now();
thread::sleep(Duration::from_millis(10));
let end = sink.now();
sink.record_timing("db.queries.select_products_ns", start, end);

// Finally, we can update a value histogram.  Technically speaking, value histograms aren't
// fundamentally different from timing histograms.  If you use a timing histogram, we do the
// math for you of getting the time difference, but other than that, identical under the hood.
let row_count = 46;
sink.record_value("db.queries.select_products_num_rows", row_count);

Scopes

Metrics can be scoped, not unlike loggers, at the Sink level. This allows sinks to easily nest themselves without callers ever needing to care about where they're located.

This feature is a simpler approach to tagging: while not as semantically rich, it provides the level of detail necessary to distinguish a single metric between multiple callsites.

For example, after getting a Sink from the Receiver, we can easily nest ourselves under the root scope and then send some metrics:

// This sink has no scope aka the root scope.  The metric will just end up as "widgets".
let mut root_sink = receiver.sink();
root_sink.increment_counter("widgets", 42);

// This sink is under the "secret" scope.  Since we derived ourselves from the root scope,
// we're not nested under anything, but our metric name will end up being "secret.widgets".
let mut scoped_sink = root_sink.scoped("secret");
scoped_sink.increment_counter("widgets", 42);

// This sink is under the "supersecret" scope, but we're also nested!  The metric name for this
// sample will end up being "secret.supersecret.widget".
let mut scoped_sink_two = scoped_sink.scoped("supersecret");
scoped_sink_two.increment_counter("widgets", 42);

// Sinks retain their scope even when cloned, so the metric name will be the same as above.
let mut cloned_sink = scoped_sink_two.clone();
cloned_sink.increment_counter("widgets", 42);

// This sink will be nested two levels deeper than its parent by using a slightly different
// input scope: scope can be a single string, or multiple strings, which is interpreted as
// nesting N levels deep.
//
// This metric name will end up being "super.secret.ultra.special.widgets".
let mut scoped_sink_three = scoped_sink.scoped(&["super", "secret", "ultra", "special"]);
scoped_sink_two.increment_counter("widgets", 42);

Labels

On top of scope support, metrics can also have labels. If scopes are for organizing metrics in a hierarchy, then labels are for differentiating the same metric being emitted from multiple sources.

This is most easily demonstrated with an example:

// We might have a function that interacts with a database and returns the number of rows it
// touched in doing so.
fn process_query(query: &str) -> u64 {
    run_query(query)
}

// We might call this function multiple times, but hitting different tables.
let rows_a = process_query("UPDATE posts SET public = 1 WHERE public = 0");
let rows_b = process_query("UPDATE comments SET public = 1 WHERE public = 0");

// Now, we want to track a metric that shows how many rows are updated overall, so the metric
// name should be the same no matter which table we update, but we'd also like to be able to
// differentiate by table, too!
sink.record_value_with_labels("db.rows_updated", rows_a, &[("table", "posts")]);
sink.record_value_with_labels("db.rows_updated", rows_b, &[("table", "comments")]);

// If you want to send a specific set of labels with every metric from this sink, you can also
// add default labels.  This action is additive, so you can call it multiple times to build up
// the set of labels sent with metrics, and labels are inherited when creating a scoped sink or
// cloning an existing sink, which allows label usage to either supplement scopes or to
// potentially replace them entirely.
sink.add_default_labels(&[("database", "primary")]);

As shown in the example, labels allow a user to submit values to the underlying metric name, while also differentiating between unique situations, whatever the facet that the user decides to utilize.

Naturally, these methods can be slightly cumbersome and visually detracting, in which case you can utilize the metric handles -- Counter, Gauge, and Histogram -- and create them with labels ahead of time.

These handles are bound to the given metric type, as well as the name, labels, and scope of the sink. Thus, there is no overhead of looking up the metric as with the record_* methods, and the values can be updated directly, and with less overhead, resulting in faster method calls.

// Let's create a counter.
let egg_count = sink.counter("eggs");

// I want a baker's dozen of eggs!
egg_count.increment();
egg_count.record(12);

// This updates the same metric as above!  We have so many eggs now!
sink.increment_counter("eggs", 12);

// Gauges and histograms don't have any extra helper methods, just `record`:
let gauge = sink.gauge("population");
gauge.record(8_000_000_000);

let histogram = sink.histogram("distribution");

// You can record a histogram value directly:
histogram.record_value(42);

// Or handily pass it two [`Delta`]-compatible values, and have it calculate the delta for you:
let start = Instant::now();
let end = Instant::now();
histogram.record_timing(start, end);

// Each of these methods also has a labels-aware companion:
let labeled_counter = sink.counter_with_labels("egg_count", &[("type", "large_brown")]);
let labeled_gauge = sink.gauge_with_labels("population", &[("country", "austria")]);
let labeled_histogram = sink.histogram_with_labels("distribution", &[("type", "performance")]);

Proxies

Sometimes, you may have a need to pull in "external" metrics: values related to your application that your application itself doesn't generate, such as system-level metrics.

Sink allows you to register a "proxy metric", which gives the ability to return metrics on-demand when a snapshot is being taken. Users provide a closure that is run every time a snapshot is being taken, which can return multiple metrics, which are then added to overall list of metrics being held by metrics-runtime itself.

If metrics are relatively expensive to calculate -- say, accessing the /proc filesytem on Linux -- then this can be a great alternative to polling them yourself and having to update them normally on some sort of schedule.

// A proxy is now registered under the name "load_stats", which is prepended to all the metrics
// generated by the closure i.e. "load_stats.avg_1min".  These metrics are also still scoped
// normally based on the [`Sink`].
sink.proxy("load_stat", || {
    let mut values = Vec::new();
    values.push((Key::from_name("avg_1min"), Measurement::Gauge(19)));
    values.push((Key::from_name("avg_5min"), Measurement::Gauge(12)));
    values.push((Key::from_name("avg_10min"), Measurement::Gauge(10)));
    values
});

Snapshots

Naturally, we need a way to get the metrics out of the system, which is where snapshots come into play. By utilizing a Controller, we can take a snapshot of the current metrics in the registry, and then output them to any desired system/interface by utilizing Observer. A number of pre-baked observers (which only concern themselves with formatting the data) and exporters (which take the formatted data and either serve it up, such as exposing an HTTP endpoint, or write it somewhere, like stdout) are available, some of which are exposed by this crate.

Let's take an example of writing out our metrics in a yaml-like format, writing them via log!:

use metrics_runtime::{
    Receiver, observers::YamlBuilder, exporters::LogExporter,
};
use log::Level;
use std::{thread, time::Duration};
let receiver = Receiver::builder().build().expect("failed to create receiver");
let mut sink = receiver.sink();

// We can update a counter.  Counters are monotonic, unsigned integers that start at 0 and
// increase over time.
// Take some measurements, similar to what we had in other examples:
sink.increment_counter("widgets", 5);
sink.update_gauge("red_balloons", 99);

let start = sink.now();
thread::sleep(Duration::from_millis(10));
let end = sink.now();
sink.record_timing("db.queries.select_products_ns", start, end);
sink.record_timing("db.gizmo_query", start, end);

let num_rows = 46;
sink.record_value("db.queries.select_products_num_rows", num_rows);

// Now create our exporter/observer configuration, and wire it up.
let exporter = LogExporter::new(
    receiver.controller(),
    YamlBuilder::new(),
    Level::Info,
    Duration::from_secs(5),
);

// This exporter will now run every 5 seconds, taking a snapshot, rendering it, and writing it
// via `log!` at the informational level. This particular exporter is running directly on the
// current thread, and not on a background thread.
//
// exporter.run();

Most exporters have the ability to run on the current thread or to be converted into a future which can be spawned on any Tokio-compatible runtime.

Facade

metrics-runtime is metrics compatible, and can be installed as the global metrics facade:

extern crate ckb_metrics_runtime as metrics_runtime;
use metrics_runtime::Receiver;

Receiver::builder()
    .build()
    .expect("failed to create receiver")
    .install();

counter!("items_processed", 42);

Modules

data	Core data types for metrics.
exporters	Commonly used exporters.
observers	Commonly used observers.

Structs

Builder	Builder for Receiver.
Controller	Handle for acquiring snapshots.
Receiver	Central store for metrics.
Sink	Handle for sending metric samples.

Enums

BuilderError	Errors during receiver creation.
Measurement	A point-in-time metric measurement.
SinkError	Errors during sink creation.

Traits

AsScoped	A value that can be used as a metric scope.
Delta	Trait for types that represent time and can be subtracted from each other to generate a delta.