@feltdb/core 0.4.12 → 0.4.14

package/dist/create/server-source/crates/feltdb/src/analytics.rs

@@ -0,0 +1,475 @@
+use serde::{Deserialize, Serialize};
+use std::collections::HashMap;
+
+#[derive(Debug, Clone, Serialize, Deserialize)]
+pub struct QueryTrace {
+    pub timestamp: u64,
+    pub fields: Vec<String>,
+    pub operator: String,
+    pub index_used: Option<String>,
+    pub execution_time_ms: f64,
+    pub rows_scanned: u64,
+    pub result_count: u64,
+}
+
+#[derive(Debug, Clone, Serialize, Deserialize)]
+pub struct IndexMetrics {
+    pub name: String,
+    pub reads: u64,
+    pub writes: u64,
+    pub hit_rate: f64,
+    pub avg_query_time_ms: f64,
+    pub avg_write_time_ms: f64,
+    pub entries: usize,
+    pub last_used_at: u64,
+    pub created_at: u64,
+}
+
+#[derive(Debug, Clone, Serialize, Deserialize)]
+pub struct TuningRecommendation {
+    pub rec_type: String, // "create_index", "drop_index", "optimize_index"
+    pub priority: String, // "high", "medium", "low"
+    pub index_name: Option<String>,
+    pub reason: String,
+    pub expected_speedup: Option<f64>,
+    pub estimated_roi: Option<f64>,
+}
+
+#[derive(Debug, Clone, Serialize, Deserialize)]
+pub struct PerformanceReport {
+    pub collection_name: String,
+    pub period: String, // "hourly", "daily", "weekly"
+    pub total_queries: usize,
+    pub indexed_queries: usize,
+    pub full_scans: usize,
+    pub avg_query_time_ms: f64,
+    pub slowest_query: Option<QueryTrace>,
+    pub index_metrics: Vec<IndexMetrics>,
+    pub recommendations: Vec<TuningRecommendation>,
+    pub timestamp: u64,
+}
+
+pub struct IndexAnalytics {
+    metrics: HashMap<String, IndexMetrics>,
+    query_traces: Vec<QueryTrace>,
+    field_query_count: HashMap<String, u64>,
+    max_traces: usize,
+}
+
+impl IndexAnalytics {
+    pub fn new() -> Self {
+        IndexAnalytics {
+            metrics: HashMap::new(),
+            query_traces: Vec::new(),
+            field_query_count: HashMap::new(),
+            max_traces: 10000,
+        }
+    }
+
+    pub fn record_index_usage(&mut self, index_name: &str, operation: &str, time_ms: f64) {
+        let metric = self
+            .metrics
+            .entry(index_name.to_string())
+            .or_insert_with(|| IndexMetrics {
+                name: index_name.to_string(),
+                reads: 0,
+                writes: 0,
+                hit_rate: 0.0,
+                avg_query_time_ms: 0.0,
+                avg_write_time_ms: 0.0,
+                entries: 0,
+                last_used_at: current_time_ms(),
+                created_at: current_time_ms(),
+            });
+
+        if operation == "read" {
+            let old_avg = metric.avg_query_time_ms;
+            metric.reads += 1;
+            metric.avg_query_time_ms = (old_avg * (metric.reads - 1) as f64 + time_ms) / metric.reads as f64;
+        } else {
+            let old_avg = metric.avg_write_time_ms;
+            metric.writes += 1;
+            metric.avg_write_time_ms = (old_avg * (metric.writes - 1) as f64 + time_ms) / metric.writes as f64;
+        }
+
+        metric.last_used_at = current_time_ms();
+    }
+
+    pub fn record_query(
+        &mut self,
+        fields: Vec<String>,
+        operator: String,
+        execution_time_ms: f64,
+        rows_scanned: u64,
+        result_count: u64,
+        index_used: Option<String>,
+    ) {
+        let trace = QueryTrace {
+            timestamp: current_time_ms(),
+            fields: fields.clone(),
+            operator,
+            index_used: index_used.clone(),
+            execution_time_ms,
+            rows_scanned,
+            result_count,
+        };
+
+        self.query_traces.push(trace);
+        if self.query_traces.len() > self.max_traces {
+            self.query_traces.remove(0);
+        }
+
+        // Track field access patterns
+        for field in &fields {
+            *self.field_query_count.entry(field.clone()).or_insert(0) += 1;
+        }
+
+        // Update hit rate for used index
+        if let Some(ref index_name) = index_used {
+            if let Some(metric) = self.metrics.get_mut(index_name) {
+                metric.reads += 1;
+            }
+        }
+    }
+
+    pub fn get_index_metrics(&self, index_name: &str) -> Option<IndexMetrics> {
+        self.metrics.get(index_name).cloned()
+    }
+
+    pub fn get_all_metrics(&self) -> Vec<IndexMetrics> {
+        self.metrics.values().cloned().collect()
+    }
+
+    pub fn get_slowest_queries(&self, limit: usize) -> Vec<QueryTrace> {
+        let mut traces = self.query_traces.clone();
+        traces.sort_by(|a, b| b.execution_time_ms.partial_cmp(&a.execution_time_ms).unwrap_or(std::cmp::Ordering::Equal));
+        traces.truncate(limit);
+        traces
+    }
+
+    pub fn get_frequent_fields(&self, limit: usize) -> Vec<(String, u64)> {
+        let mut fields: Vec<_> = self.field_query_count.iter().map(|(k, v)| (k.clone(), *v)).collect();
+        fields.sort_by(|a, b| b.1.cmp(&a.1));
+        fields.truncate(limit);
+        fields
+    }
+
+    fn calculate_hit_rate(&self, index_name: &str) -> f64 {
+        let metric = match self.metrics.get(index_name) {
+            Some(m) => m,
+            None => return 0.0,
+        };
+
+        if metric.reads == 0 {
+            return 0.0;
+        }
+
+        if self.query_traces.is_empty() {
+            return 0.0;
+        }
+
+        let queries_using_index = self
+            .query_traces
+            .iter()
+            .filter(|q| q.index_used.as_ref().map_or(false, |u| u == index_name))
+            .count();
+
+        let total_queries = self.query_traces.len();
+        if total_queries == 0 {
+            0.0
+        } else {
+            queries_using_index as f64 / total_queries as f64
+        }
+    }
+
+    pub fn generate_report(
+        &self,
+        collection_name: &str,
+        period: &str,
+    ) -> PerformanceReport {
+        let now = current_time_ms();
+        let cutoff = match period {
+            "hourly" => now - 3_600_000,
+            "weekly" => now - 604_800_000,
+            _ => now - 86_400_000, // daily
+        };
+
+        let recent_traces: Vec<_> = self
+            .query_traces
+            .iter()
+            .filter(|t| t.timestamp > cutoff)
+            .cloned()
+            .collect();
+
+        let indexed_queries = recent_traces.iter().filter(|t| t.index_used.is_some()).count();
+        let full_scans = recent_traces.len() - indexed_queries;
+
+        let avg_query_time = if recent_traces.is_empty() {
+            0.0
+        } else {
+            recent_traces.iter().map(|t| t.execution_time_ms).sum::<f64>() / recent_traces.len() as f64
+        };
+
+        let slowest_query = if recent_traces.is_empty() {
+            None
+        } else {
+            let mut traces = recent_traces.clone();
+            traces.sort_by(|a, b| b.execution_time_ms.partial_cmp(&a.execution_time_ms).unwrap_or(std::cmp::Ordering::Equal));
+            traces.first().cloned()
+        };
+
+        let mut metrics = self.get_all_metrics();
+        for metric in &mut metrics {
+            metric.hit_rate = self.calculate_hit_rate(&metric.name);
+        }
+
+        let recommendations = self.generate_recommendations(&recent_traces, &metrics);
+
+        PerformanceReport {
+            collection_name: collection_name.to_string(),
+            period: period.to_string(),
+            total_queries: recent_traces.len(),
+            indexed_queries,
+            full_scans,
+            avg_query_time_ms: avg_query_time,
+            slowest_query,
+            index_metrics: metrics,
+            recommendations,
+            timestamp: now,
+        }
+    }
+
+    fn generate_recommendations(
+        &self,
+        traces: &[QueryTrace],
+        metrics: &[IndexMetrics],
+    ) -> Vec<TuningRecommendation> {
+        let mut recommendations = Vec::new();
+
+        if traces.is_empty() {
+            return recommendations;
+        }
+
+        // Recommend indexes for frequently scanned fields
+        let mut field_scans: HashMap<String, u64> = HashMap::new();
+        for trace in traces {
+            if trace.index_used.is_none() && trace.rows_scanned > 1000 {
+                for field in &trace.fields {
+                    *field_scans.entry(field.clone()).or_insert(0) += trace.rows_scanned;
+                }
+            }
+        }
+
+        for (field, total_rows_scanned) in field_scans {
+            let avg_scanned_per_query = total_rows_scanned as f64 / traces.len() as f64;
+            if avg_scanned_per_query > 1000.0 {
+                recommendations.push(TuningRecommendation {
+                    rec_type: "create_index".to_string(),
+                    priority: "high".to_string(),
+                    index_name: None,
+                    reason: format!("Field '{}' scanned {:.0} rows per query", field, avg_scanned_per_query),
+                    expected_speedup: Some(avg_scanned_per_query / 10.0),
+                    estimated_roi: None,
+                });
+            }
+        }
+
+        // Recommend dropping unused indexes
+        let unused_threshold = 0.01;
+        let now = current_time_ms();
+        for metric in metrics {
+            if metric.hit_rate < unused_threshold && (now - metric.last_used_at) > 86_400_000 {
+                recommendations.push(TuningRecommendation {
+                    rec_type: "drop_index".to_string(),
+                    priority: "low".to_string(),
+                    index_name: Some(metric.name.clone()),
+                    reason: format!("Index has {:.1}% hit rate and unused for 24h", metric.hit_rate * 100.0),
+                    expected_speedup: None,
+                    estimated_roi: Some(-0.01),
+                });
+            }
+        }
+
+        // Recommend compound indexes
+        let mut field_pairs: HashMap<String, u64> = HashMap::new();
+        for trace in traces {
+            if trace.fields.len() == 2 {
+                let mut sorted = trace.fields.clone();
+                sorted.sort();
+                let key = sorted.join("|");
+                *field_pairs.entry(key).or_insert(0) += 1;
+            }
+        }
+
+        for (fields, count) in field_pairs {
+            if (count as f64) > traces.len() as f64 * 0.1 {
+                recommendations.push(TuningRecommendation {
+                    rec_type: "create_index".to_string(),
+                    priority: "medium".to_string(),
+                    index_name: None,
+                    reason: format!("Fields '{}' queried together in {} queries", fields.replace('|', ", "), count),
+                    expected_speedup: Some(5.0),
+                    estimated_roi: None,
+                });
+            }
+        }
+
+        // Sort by priority
+        let priority_order = |p: &str| match p {
+            "high" => 0,
+            "medium" => 1,
+            _ => 2,
+        };
+
+        recommendations.sort_by(|a, b| {
+            let priority_cmp = priority_order(&a.priority).cmp(&priority_order(&b.priority));
+            if priority_cmp == std::cmp::Ordering::Equal {
+                let speedup_a = a.expected_speedup.unwrap_or(0.0);
+                let speedup_b = b.expected_speedup.unwrap_or(0.0);
+                speedup_b.partial_cmp(&speedup_a).unwrap_or(std::cmp::Ordering::Equal)
+            } else {
+                priority_cmp
+            }
+        });
+
+        recommendations.truncate(5);
+        recommendations
+    }
+
+    pub fn reset(&mut self) {
+        self.metrics.clear();
+        self.query_traces.clear();
+        self.field_query_count.clear();
+    }
+}
+
+fn current_time_ms() -> u64 {
+    std::time::SystemTime::now()
+        .duration_since(std::time::UNIX_EPOCH)
+        .map(|d| d.as_millis() as u64)
+        .unwrap_or(0)
+}
+
+/// Automatic tuning engine - applies recommendations automatically.
+pub struct AutoTuner {
+    applied_recommendations: std::collections::HashSet<String>,
+}
+
+impl AutoTuner {
+    pub fn new(_analytics: IndexAnalytics) -> Self {
+        AutoTuner {
+            applied_recommendations: std::collections::HashSet::new(),
+        }
+    }
+
+    /// Check if a recommendation should be auto-applied.
+    pub fn should_auto_apply(&self, recommendation: &TuningRecommendation) -> bool {
+        if recommendation.priority != "high" {
+            return false;
+        }
+
+        let key = format!("{}:{}", recommendation.rec_type, recommendation.index_name.as_ref().unwrap_or(&recommendation.reason));
+        if self.applied_recommendations.contains(&key) {
+            return false;
+        }
+
+        if (recommendation.expected_speedup.unwrap_or(0.0)) < 5.0 {
+            return false;
+        }
+
+        true
+    }
+
+    /// Apply a recommendation.
+    pub fn apply_recommendation(&mut self, recommendation: &TuningRecommendation) -> bool {
+        if !self.should_auto_apply(recommendation) {
+            return false;
+        }
+
+        let key = format!("{}:{}", recommendation.rec_type, recommendation.index_name.as_ref().unwrap_or(&recommendation.reason));
+        self.applied_recommendations.insert(key);
+        true
+    }
+
+    /// Get recommendations that are ready for auto-apply.
+    pub fn get_auto_applicable_recommendations(&self, report: &PerformanceReport) -> Vec<TuningRecommendation> {
+        report
+            .recommendations
+            .iter()
+            .filter(|r| self.should_auto_apply(r))
+            .cloned()
+            .collect()
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn test_record_index_usage() {
+        let mut analytics = IndexAnalytics::new();
+        analytics.record_index_usage("test_idx", "read", 2.5);
+
+        let metrics = analytics.get_index_metrics("test_idx");
+        assert!(metrics.is_some());
+        assert_eq!(metrics.unwrap().reads, 1);
+    }
+
+    #[test]
+    fn test_record_query() {
+        let mut analytics = IndexAnalytics::new();
+        analytics.record_query(
+            vec!["status".to_string()],
+            "eq".to_string(),
+            1.5,
+            100,
+            50,
+            Some("status_idx".to_string()),
+        );
+
+        assert_eq!(analytics.query_traces.len(), 1);
+        assert_eq!(analytics.field_query_count.get("status"), Some(&1));
+    }
+
+    #[test]
+    fn test_slowest_queries() {
+        let mut analytics = IndexAnalytics::new();
+        analytics.record_query(vec!["a".to_string()], "eq".to_string(), 1.0, 10, 5, None);
+        analytics.record_query(vec!["b".to_string()], "eq".to_string(), 5.0, 20, 10, None);
+        analytics.record_query(vec!["c".to_string()], "eq".to_string(), 3.0, 15, 8, None);
+
+        let slowest = analytics.get_slowest_queries(2);
+        assert_eq!(slowest.len(), 2);
+        assert_eq!(slowest[0].execution_time_ms, 5.0);
+        assert_eq!(slowest[1].execution_time_ms, 3.0);
+    }
+
+    #[test]
+    fn test_generate_report() {
+        let mut analytics = IndexAnalytics::new();
+        analytics.record_query(
+            vec!["status".to_string()],
+            "eq".to_string(),
+            2.5,
+            100,
+            50,
+            Some("status_idx".to_string()),
+        );
+        analytics.record_query(
+            vec!["created".to_string()],
+            "range".to_string(),
+            50.0,
+            10000,
+            100,
+            None,
+        );
+
+        let report = analytics.generate_report("orders", "daily");
+        assert_eq!(report.collection_name, "orders");
+        assert_eq!(report.total_queries, 2);
+        assert_eq!(report.indexed_queries, 1);
+        assert_eq!(report.full_scans, 1);
+        assert!(report.recommendations.len() > 0);
+    }
+}