Struct AnalyticsConfig 

pub struct AnalyticsConfig {
    pub path: String,
    pub flush_ms: Option<u64>,
    pub retention: Option<AnalyticsRetention>,
    pub dimensions: Vec<AnalyticsDimension>,
    pub exports: Option<AnalyticsExports>,
    pub live_sketches: Option<AnalyticsLiveSketches>,
    pub insights: Option<AnalyticsInsights>,
    pub replay_capture: Option<AnalyticsReplayCapture>,
    pub sampled_degradation: Option<AnalyticsSampledDegradation>,
}

The embedded analytics store config (DW-043, gateway.analytics).

Fields

path: String

Filesystem path of the analytics SQLite database. The file is created (and migrated) on first open. A separate file from the state store on purpose: retention deletes and incremental vacuum churn must never compact the identity store.

flush_ms: Option<u64>

Maximum batch latency of the background writer, in milliseconds (default 1000; validated to [100, 60 000]). Not a correctness knob — the rollup grace covers multiples of this.

retention: Option<AnalyticsRetention>

Per-store retention, in milliseconds (defaults: raw 24 h, 1m 48 h, 5m 7 d, 1h 30 d, 1d 365 d). Validation enforces each floor (a granularity must outlive its own window comfortably) and monotonicity (no coarser table may expire before a finer one).

dimensions: Vec<AnalyticsDimension>

Custom dimensions (DW-043): request-header-sourced tags that become analytics dimensions — x-plan as dimension plan, for consumer tier / plan / feature-flag style cuts the fixed dimensions cannot express. Extracted at request completion capture; analytics-only (never added to the access log). At most 16 (cardinality guard); values are capped at 128 bytes (longer values are skipped for that request).

exports: Option<AnalyticsExports>

Scheduled usage-report exports (DW-120): durable CSV/JSON dumps of the per-consumer usage statement, one file set per closed UTC calendar window, written by a background worker that reuses the analytics store’s scheduling machinery. Absent (the default): no exports run and the admin export endpoints answer that exports are not configured. See the analytics::exports module docs for the schedule, the statement’s reconciliation contract with the query API, and the quota-column window-alignment rule.

live_sketches: Option<AnalyticsLiveSketches>

Live in-process sketches (DW-092): sub-second-freshness aggregation maintained synchronously on the request-completion hot path (per-route rolling window with counts, errors, and latency samples). Never blocks the dataplane — the sketch is updated in place under atomics and a short mutex over a capped sample vector. Absent (the default): no live sketches; the GET /analytics/live endpoint answers analytics_not_configured.

insights: Option<AnalyticsInsights>

ML traffic insights (DW-092): EWMA-based capacity forecasting and seasonal-baseline anomaly detection over the live sketch window rotations. The insights engine is in-process and hand-rolled (no ML crates): a minute-of-day ring buffer holds the seasonal pattern, an EWMA over recent windows carries the trend, and anomaly detection compares the current window’s shape to the seasonal baseline. Absent (the default): no forecasting or anomaly detection; the GET /analytics/forecast and GET /analytics/anomalies endpoints answer analytics_not_configured.

replay_capture: Option<AnalyticsReplayCapture>

Replay time-travel debugging capture (DW-102): opt-in capture of request detail (redacted headers, auth identity) into the raw analytics table for offline replay against candidate configs. Absent (the default): no replay capture; the raw table’s request_headers_redacted and auth_identity columns stay NULL. When present with enabled: true, the fire-and-forget analytics writer records the redacted headers and auth identity alongside the existing access fields, bounded by retain_hours. See AnalyticsReplayCapture.

sampled_degradation: Option<AnalyticsSampledDegradation>

REL-10 (#221): sampled degradation under channel pressure. When the analytics channel’s fill ratio exceeds high_watermark, the offer path probabilistically drops records (keeping 1 in sample_rate) BEFORE the channel fills, so a sustained burst degrades to sampled telemetry rather than a hard drop cliff. Below low_watermark every record is kept. Defaults to off (no sampling); when enabled, high_watermark must be > low_watermark and both are in (0.0, 1.0].

Trait Implementations

impl Clone for AnalyticsConfig

fn clone(&self) -> AnalyticsConfig

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for AnalyticsConfig

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

Formats the value using the given formatter.

impl<'de> Deserialize<'de> for AnalyticsConfig

fn deserialize<__D>(__deserializer: __D) -> Result<Self, __D::Error>

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl JsonSchema for AnalyticsConfig

fn schema_name() -> Cow<'static, str>

The name of the generated JSON Schema.

fn schema_id() -> Cow<'static, str>

Returns a string that uniquely identifies the schema produced by this type.

fn json_schema(generator: &mut SchemaGenerator) -> Schema

Generates a JSON Schema for this type.

fn inline_schema() -> bool

Whether JSON Schemas generated for this type should be included directly in parent schemas, rather than being re-used where possible using the $ref keyword.

impl PartialEq for AnalyticsConfig

fn eq(&self, other: &AnalyticsConfig) -> bool

Tests for self and other values to be equal, and is used by ==.

fn ne(&self, other: &Rhs) -> bool

Tests for !=. The default implementation is almost always sufficient, and should not be overridden without very good reason.

impl Serialize for AnalyticsConfig

fn serialize<__S>(&self, __serializer: __S) -> Result<__S::Ok, __S::Error>

where __S: Serializer,

Serialize this value into the given Serde serializer.

impl StructuralPartialEq for AnalyticsConfig