Cachee Helm Charts

Official Kubernetes Helm charts for deploying Cachee.ai proxy — sub-microsecond caching for any Redis-compatible workload.

Charts

Chart	Description	Status
cachee-proxy	Standalone Cachee proxy deployment	Stable
cachee-sidecar	Sidecar injector for existing pods	Stable
cachee-overlay	Overlay mode — accelerate existing Redis/ElastiCache	Stable

Quick Start

helm repo add cachee https://charts.cachee.ai
helm repo update

# Deploy Cachee proxy
helm install cachee cachee/cachee-proxy \
  --set upstream.host=your-redis.example.com \
  --set upstream.port=6379 \
  --set l1.maxKeys=1000000

Architecture

┌─────────────┐     ┌──────────────────────┐     ┌─────────────┐
│  Your App   │────▶│   Cachee Proxy (L1)  │────▶│ Redis / EC  │
│             │     │   17ns cache reads    │     │   (L2/L3)   │
└─────────────┘     └──────────────────────┘     └─────────────┘
     :6380                                            :6379

• L1 (Cachee): Hot keys in application memory — 17ns reads
• L2 (Redis): Warm keys — automatic fallthrough on L1 miss
• L3 (Origin): Cold keys — your existing backing store

How Cachee works | Performance benchmarks

Deployment Modes

Sidecar (Recommended for Kubernetes)

Cachee runs as a sidecar in each pod. Zero network hops for cache reads.

# values-sidecar.yaml
mode: sidecar
resources:
  requests:
    memory: "256Mi"
    cpu: "100m"
  limits:
    memory: "1Gi"
    cpu: "500m"
upstream:
  host: redis-cluster.default.svc
  port: 6379
l1:
  maxKeys: 500000
  ttlSeconds: 300

Overlay (Accelerate Existing Cache)

Point Cachee at your existing ElastiCache, Redis Cloud, Azure Cache, GCP Memorystore, or Upstash instance.

# values-overlay.yaml
mode: overlay
upstream:
  host: my-cluster.xxxxx.use1.cache.amazonaws.com
  port: 6379
  tls: true

Supported providers

Standalone

Deploy Cachee as a separate service for shared caching across multiple applications.

# values-standalone.yaml
mode: standalone
replicas: 3
service:
  type: ClusterIP
  port: 6380

Configuration Reference

Parameter	Description	Default
mode	Deployment mode: sidecar, overlay, standalone	sidecar
upstream.host	Redis/cache backend hostname	redis
upstream.port	Backend port	6379
upstream.tls	Enable TLS to upstream	false
l1.maxKeys	Maximum L1 cache entries	1000000
l1.ttlSeconds	L1 TTL in seconds	300
health.port	Health check port	9090
metrics.enabled	Enable Prometheus metrics	true
metrics.port	Metrics port	9091

Monitoring

Cachee exposes Prometheus metrics on the configured metrics port:

• cachee_l1_hits_total — L1 cache hits
• cachee_l1_misses_total — L1 cache misses
• cachee_l2_hits_total — L2 (upstream) hits
• cachee_latency_nanoseconds — Read latency histogram
• cachee_throughput_ops — Operations per second

Links

• Website: cachee.ai
• Deployment Guide: cachee.ai/how-it-works
• Installation: cachee.ai/install
• Integrations: cachee.ai/integrations
• Pricing: cachee.ai/pricing
• Examples: github.com/HapPhi/cachee-examples
• Benchmarks: github.com/HapPhi/cachee-benchmarks

License

Apache 2.0