hll experiment (#6312)

## Problem

Measuring cardinality using logs is expensive and slow.

## Summary of changes

Implement a pre-aggregated HyperLogLog-based cardinality estimate.
HyperLogLog estimates the cardinality of a set by using the probability
that the uniform hash of a value will have a run of n 0s at the end is
`1/2^n`, therefore, having observed a run of `n` 0s suggests we have
measured `2^n` distinct values. By using multiple shards, we can use the
harmonic mean to get a more accurate estimate.

We record this into a Prometheus time-series. HyperLogLog counts can be
merged by taking the `max` of each shard. We can apply a `max_over_time`
in order to find the estimate of cardinality of distinct values over
time

proxy/src/context.rs

@@ -91,6 +91,11 @@ impl RequestMonitoring {
 
     pub fn set_endpoint_id(&mut self, endpoint_id: Option<EndpointId>) {
         self.endpoint_id = endpoint_id.or_else(|| self.endpoint_id.clone());
+        if let Some(ep) = &self.endpoint_id {
+            crate::metrics::CONNECTING_ENDPOINTS
+                .with_label_values(&[self.protocol])
+                .measure(&ep);
+        }
     }
 
     pub fn set_application(&mut self, app: Option<SmolStr>) {

proxy/src/metrics.rs

@@ -1,10 +1,7 @@
 use ::metrics::{
-    exponential_buckets, register_int_counter_pair_vec, register_int_counter_vec,
-    IntCounterPairVec, IntCounterVec,
-};
-use prometheus::{
-    register_histogram, register_histogram_vec, register_int_gauge_vec, Histogram, HistogramVec,
-    IntGaugeVec,
+    exponential_buckets, register_histogram, register_histogram_vec, register_hll_vec,
+    register_int_counter_pair_vec, register_int_counter_vec, register_int_gauge_vec, Histogram,
+    HistogramVec, HyperLogLogVec, IntCounterPairVec, IntCounterVec, IntGaugeVec,
 };
 
 use once_cell::sync::Lazy;
@@ -236,3 +233,13 @@ pub const fn bool_to_str(x: bool) -> &'static str {
         "false"
     }
 }
+
+pub static CONNECTING_ENDPOINTS: Lazy<HyperLogLogVec<32>> = Lazy::new(|| {
+    register_hll_vec!(
+        32,
+        "proxy_connecting_endpoints",
+        "HLL approximate cardinality of endpoints that are connecting",
+        &["protocol"],
+    )
+    .unwrap()
+});