BitzOrcas currently uses IMemoryCache for single-instance in-memory caching needs. A multi-tier caching strategy with FusionCache is on the roadmap for production multi-instance deployments.

Current usage

Idempotency deduplication

The IdempotencyPipelineBehavior uses IMemoryCache to deduplicate requests carrying an Idempotency-Key header:

// Request with Idempotency-Key: "req-abc123"
// → First call: executes handler, caches result
// → Duplicate call: returns cached result immediately

Registered in ApiPipelineRegistration:

services.AddMemoryCache(); // Powers IdempotencyPipelineBehavior

Rate limiting counters

ASP.NET Core’s built-in RateLimiter uses in-memory partitioned counters. See Web building block for policy details.

Roadmap: FusionCache

Planned multi-tier caching with FusionCache:

┌───────────────────────────────────────┐
│           Application Layer           │
│               Cache Get/Set           │
└───────────────┬───────────────────────┘
                ▼
┌───────────────────────────────────────┐
│           FusionCache                 │
│  ┌─────────────┐  ┌───────────────┐  │
│  │  L1 Memory  │  │  L2 Redis/    │  │
│  │  (local)    │→ │  Distributed  │  │
│  └─────────────┘  └───────────────┘  │
│  + Stampede protection                │
│  + Fail-safe fallback                 │
│  + OpenTelemetry instrumentation     │
└───────────────────────────────────────┘

Planned features

L1/L2 cache tier — in-process memory + distributed (Redis/Valkey)
Stampede protection — prevent cache thundering herd
Fail-safe — graceful fallback when L2 is unavailable
OTel instrumentation — cache hit/miss metrics
Cache key versioning — automatic invalidation on schema changes

Configuration (future)

{
  "FusionCache": {
    "L1Enabled": true,
    "L2Enabled": true,
    "L2ConnectionString": "localhost:6379",
    "DefaultDuration": "00:05:00",
    "MaximumDuration": "01:00:00"
  }
}