brainy/src/utils/adaptiveBackpressure.ts
David Snelling 9842af2fe3 feat: add automatic adaptive performance optimization for high-volume scenarios
- Implement AdaptiveSocketManager for zero-config socket pool scaling
- Add AdaptiveBackpressure for intelligent flow control with circuit breaker
- Create PerformanceMonitor for real-time metrics and auto-optimization
- Automatically adapt to load patterns without manual configuration
- Self-healing system that learns from usage patterns
- Dynamically adjust batch sizes based on system resources
- Automatic recovery from socket exhaustion scenarios
- No configuration required - system adapts automatically

This addresses socket exhaustion issues reported by bluesky-package
by providing automatic, adaptive resource management that scales
based on actual load patterns.
2025-08-07 08:37:15 -07:00

443 lines
No EOL
12 KiB
TypeScript

/**
* Adaptive Backpressure System
* Automatically manages request flow and prevents system overload
* Self-healing with pattern learning for optimal throughput
*/
import { createModuleLogger } from './logger.js'
interface BackpressureMetrics {
queueDepth: number
processingRate: number
errorRate: number
latency: number
throughput: number
}
interface BackpressureConfig {
maxQueueDepth: number
targetLatency: number
minThroughput: number
adaptationRate: number
}
/**
* Self-healing backpressure manager that learns from load patterns
*/
export class AdaptiveBackpressure {
private logger = createModuleLogger('AdaptiveBackpressure')
// Queue management
private queue: Array<{
id: string
priority: number
timestamp: number
resolve: () => void
}> = []
// Active operations tracking
private activeOperations = new Set<string>()
private maxConcurrent = 100
// Metrics tracking
private metrics: BackpressureMetrics = {
queueDepth: 0,
processingRate: 0,
errorRate: 0,
latency: 0,
throughput: 0
}
// Configuration that adapts over time
private config: BackpressureConfig = {
maxQueueDepth: 1000,
targetLatency: 1000, // 1 second target
minThroughput: 10, // Minimum 10 ops/sec
adaptationRate: 0.1 // How quickly to adapt
}
// Historical patterns for learning
private patterns: Array<{
timestamp: number
load: number
optimal: number
}> = []
// Circuit breaker state
private circuitState: 'closed' | 'open' | 'half-open' = 'closed'
private circuitOpenTime = 0
private circuitFailures = 0
private circuitThreshold = 5
private circuitTimeout = 30000 // 30 seconds
// Performance tracking
private operationTimes = new Map<string, number>()
private completedOps: number[] = []
private errorOps = 0
private lastAdaptation = Date.now()
/**
* Request permission to proceed with an operation
*/
public async requestPermission(
operationId: string,
priority: number = 1
): Promise<void> {
// Check circuit breaker
if (this.isCircuitOpen()) {
throw new Error('Circuit breaker is open - system is recovering')
}
// Fast path for low load
if (this.activeOperations.size < this.maxConcurrent * 0.5 && this.queue.length === 0) {
this.activeOperations.add(operationId)
this.operationTimes.set(operationId, Date.now())
return
}
// Check if we need to queue
if (this.activeOperations.size >= this.maxConcurrent) {
// Check queue depth
if (this.queue.length >= this.config.maxQueueDepth) {
throw new Error('Backpressure queue is full - try again later')
}
// Add to queue and wait
return new Promise<void>((resolve) => {
this.queue.push({
id: operationId,
priority,
timestamp: Date.now(),
resolve
})
// Sort queue by priority (higher priority first)
this.queue.sort((a, b) => b.priority - a.priority)
// Update metrics
this.metrics.queueDepth = this.queue.length
})
}
// Add to active operations
this.activeOperations.add(operationId)
this.operationTimes.set(operationId, Date.now())
}
/**
* Release permission after operation completes
*/
public releasePermission(operationId: string, success: boolean = true): void {
// Remove from active operations
this.activeOperations.delete(operationId)
// Track completion time
const startTime = this.operationTimes.get(operationId)
if (startTime) {
const duration = Date.now() - startTime
this.completedOps.push(duration)
this.operationTimes.delete(operationId)
// Keep array bounded
if (this.completedOps.length > 1000) {
this.completedOps = this.completedOps.slice(-500)
}
}
// Track errors for circuit breaker
if (!success) {
this.errorOps++
this.circuitFailures++
// Check if we should open circuit
if (this.circuitFailures >= this.circuitThreshold) {
this.openCircuit()
}
} else {
// Reset circuit failures on success
if (this.circuitState === 'half-open') {
this.closeCircuit()
}
}
// Process queue if there are waiting operations
if (this.queue.length > 0 && this.activeOperations.size < this.maxConcurrent) {
const next = this.queue.shift()
if (next) {
this.activeOperations.add(next.id)
this.operationTimes.set(next.id, Date.now())
next.resolve()
// Update metrics
this.metrics.queueDepth = this.queue.length
}
}
// Adapt configuration periodically
this.adaptIfNeeded()
}
/**
* Check if circuit breaker is open
*/
private isCircuitOpen(): boolean {
if (this.circuitState === 'open') {
// Check if timeout has passed
if (Date.now() - this.circuitOpenTime > this.circuitTimeout) {
this.circuitState = 'half-open'
this.logger.info('Circuit breaker entering half-open state')
return false
}
return true
}
return false
}
/**
* Open the circuit breaker
*/
private openCircuit(): void {
if (this.circuitState !== 'open') {
this.circuitState = 'open'
this.circuitOpenTime = Date.now()
this.logger.warn('Circuit breaker opened due to high error rate')
// Reduce load immediately
this.maxConcurrent = Math.max(10, Math.floor(this.maxConcurrent * 0.3))
}
}
/**
* Close the circuit breaker
*/
private closeCircuit(): void {
this.circuitState = 'closed'
this.circuitFailures = 0
this.logger.info('Circuit breaker closed - system recovered')
// Gradually increase capacity
this.maxConcurrent = Math.min(500, Math.floor(this.maxConcurrent * 1.5))
}
/**
* Adapt configuration based on metrics
*/
private adaptIfNeeded(): void {
const now = Date.now()
if (now - this.lastAdaptation < 5000) { // Adapt every 5 seconds
return
}
this.lastAdaptation = now
this.updateMetrics()
// Learn from current patterns
this.learnPattern()
// Adapt based on metrics
this.adaptConfiguration()
}
/**
* Update current metrics
*/
private updateMetrics(): void {
// Calculate processing rate
this.metrics.processingRate = this.completedOps.length > 0
? 1000 / (this.completedOps.reduce((a, b) => a + b, 0) / this.completedOps.length)
: 0
// Calculate error rate
const totalOps = this.completedOps.length + this.errorOps
this.metrics.errorRate = totalOps > 0 ? this.errorOps / totalOps : 0
// Calculate average latency
this.metrics.latency = this.completedOps.length > 0
? this.completedOps.reduce((a, b) => a + b, 0) / this.completedOps.length
: 0
// Calculate throughput
this.metrics.throughput = this.activeOperations.size + this.metrics.processingRate
// Reset error counter periodically
if (this.completedOps.length > 100) {
this.errorOps = Math.floor(this.errorOps * 0.9) // Decay error count
}
}
/**
* Learn from current load patterns
*/
private learnPattern(): void {
const currentLoad = this.activeOperations.size + this.queue.length
const optimalConcurrency = this.calculateOptimalConcurrency()
this.patterns.push({
timestamp: Date.now(),
load: currentLoad,
optimal: optimalConcurrency
})
// Keep patterns bounded
if (this.patterns.length > 1000) {
this.patterns = this.patterns.slice(-500)
}
}
/**
* Calculate optimal concurrency based on Little's Law
*/
private calculateOptimalConcurrency(): number {
// Little's Law: L = λ * W
// L = number of requests in system
// λ = arrival rate
// W = average time in system
if (this.metrics.latency === 0 || this.metrics.processingRate === 0) {
return this.maxConcurrent // Keep current if no data
}
// Target: Keep latency under target while maximizing throughput
const targetConcurrency = Math.ceil(
this.metrics.processingRate * (this.config.targetLatency / 1000)
)
// Adjust based on error rate
const errorAdjustment = 1 - (this.metrics.errorRate * 2) // Reduce by up to 50% for errors
// Apply adjustment
const adjusted = Math.floor(targetConcurrency * errorAdjustment)
// Apply bounds
return Math.max(10, Math.min(500, adjusted))
}
/**
* Adapt configuration based on metrics and patterns
*/
private adaptConfiguration(): void {
const optimal = this.calculateOptimalConcurrency()
const current = this.maxConcurrent
// Smooth adaptation using exponential moving average
const newConcurrency = Math.floor(
current * (1 - this.config.adaptationRate) +
optimal * this.config.adaptationRate
)
// Check if adaptation is needed
if (Math.abs(newConcurrency - current) > current * 0.1) { // 10% threshold
const oldValue = this.maxConcurrent
this.maxConcurrent = newConcurrency
this.logger.debug('Adapted concurrency', {
from: oldValue,
to: newConcurrency,
metrics: this.metrics
})
}
// Adapt queue depth based on throughput
if (this.metrics.throughput > 0) {
// Allow queue depth to be 10 seconds worth of throughput
this.config.maxQueueDepth = Math.max(
100,
Math.min(10000, Math.floor(this.metrics.throughput * 10))
)
}
// Adapt circuit breaker threshold based on error patterns
if (this.metrics.errorRate < 0.01 && this.circuitThreshold > 5) {
this.circuitThreshold = Math.max(5, this.circuitThreshold - 1)
} else if (this.metrics.errorRate > 0.05 && this.circuitThreshold < 20) {
this.circuitThreshold = Math.min(20, this.circuitThreshold + 1)
}
}
/**
* Predict future load based on patterns
*/
public predictLoad(futureSeconds: number = 60): number {
if (this.patterns.length < 10) {
return this.maxConcurrent // Not enough data
}
// Simple linear regression on recent patterns
const recentPatterns = this.patterns.slice(-50)
const n = recentPatterns.length
// Calculate averages
let sumX = 0, sumY = 0, sumXY = 0, sumX2 = 0
const startTime = recentPatterns[0].timestamp
recentPatterns.forEach(p => {
const x = (p.timestamp - startTime) / 1000 // Time in seconds
const y = p.load
sumX += x
sumY += y
sumXY += x * y
sumX2 += x * x
})
// Calculate slope and intercept
const slope = (n * sumXY - sumX * sumY) / (n * sumX2 - sumX * sumX)
const intercept = (sumY - slope * sumX) / n
// Predict future load
const currentTime = (Date.now() - startTime) / 1000
const predictedLoad = intercept + slope * (currentTime + futureSeconds)
return Math.max(0, Math.min(this.config.maxQueueDepth, Math.floor(predictedLoad)))
}
/**
* Get current configuration and metrics
*/
public getStatus(): {
config: BackpressureConfig
metrics: BackpressureMetrics
circuit: string
maxConcurrent: number
activeOps: number
queueLength: number
} {
return {
config: { ...this.config },
metrics: { ...this.metrics },
circuit: this.circuitState,
maxConcurrent: this.maxConcurrent,
activeOps: this.activeOperations.size,
queueLength: this.queue.length
}
}
/**
* Reset to default state
*/
public reset(): void {
this.queue = []
this.activeOperations.clear()
this.operationTimes.clear()
this.completedOps = []
this.errorOps = 0
this.patterns = []
this.circuitState = 'closed'
this.circuitFailures = 0
this.maxConcurrent = 100
this.logger.info('Backpressure system reset to defaults')
}
}
// Global singleton instance
let globalBackpressure: AdaptiveBackpressure | null = null
/**
* Get the global backpressure instance
*/
export function getGlobalBackpressure(): AdaptiveBackpressure {
if (!globalBackpressure) {
globalBackpressure = new AdaptiveBackpressure()
}
return globalBackpressure
}