zae-limiter: DynamoDB hot partition throttling enables per-entity Denial of Service

Summary

All rate limit buckets for a single entity share the same DynamoDB partition key (namespace/ENTITY#{id}). A high-traffic entity can exceed DynamoDB's per-partition throughput limits (~1,000 WCU/sec), causing throttling that degrades service for that entity — and potentially co-located entities in the same partition.

Details

Each acquire() call performs a TransactWriteItems (or UpdateItem in speculative mode) against items sharing the same partition key. For cascade entities, this doubles to 2-4 writes per request (child + parent). At sustained rates above ~500 req/sec for a single entity, DynamoDB's adaptive capacity may not redistribute fast enough, causing ProvisionedThroughputExceededException.

The library has no built-in mitigation:

• No partition key sharding/salting
• No write coalescing or batching
• No client-side admission control before hitting DynamoDB
• RateLimiterUnavailable is raised but the caller has already been delayed

Impact

• Availability: High-traffic entities experience elevated latency and rejected requests beyond what their rate limits specify
• Fairness: Other entities sharing the same DynamoDB partition may experience collateral throttling
• Multi-tenant risk: In a shared LLM proxy scenario, one tenant's burst traffic could degrade service for others

Reproduction

1. Create an entity with high rate limits (e.g., 100,000 rpm)
2. Send sustained traffic at 1,000+ req/sec to a single entity
3. Observe DynamoDB ThrottledRequests CloudWatch metric increasing
4. Observe acquire() latency spikes and RateLimiterUnavailable exceptions

Remediation Design: Pre-Shard Buckets

• Move buckets to PK={ns}/BUCKET#{entity}#{resource}#{shard}, SK=#STATE — one partition per (entity, resource, shard)
• Auto-inject wcu:1000 reserved limit on every bucket — tracks DynamoDB partition write pressure in-band (name may change during implementation)
• Shard doubling (1→2→4→8) triggered by client on wcu exhaustion or proactively by aggregator
• Shard 0 at suffix #0 is source of truth for shard_count. Aggregator propagates to other shards
• Original limits stored on bucket, effective limits derived: . Infrastructure limits () not divided