Sheet 06

Backend Microservices Fleet

Service instance sizing, core/memory allocation, throughput capacity, and cluster headroom buffer.

Cluster Headroom1018% — OK
Total Pod Instances
29 Pods
Across 6 services
Total Compute Cores
110 vCPUs
Allocated CPU
Total Cluster Memory
244 GB RAM
Working memory
Aggregate RPS Capacity
74,500 req/s
Peak demand: 6,667

Service Deployment Matrix

Adjust instance counts below to observe live headroom changes
Service NameInstancesCores / PodRAM / PodRPS / PodTotal RPS CapArchitecture Role & Notes
API Gateway
948 GB3,00027,000 req/sRate limiting, JWT auth, TLS, edge routing
Order Service
48 GB2,00012,000 req/sOutbox pattern, order saga, Kafka producer
Payment Service
48 GB1,5006,000 req/sIdempotency keys, PCI-DSS scoped zone
Inventory Service
48 GB2,50010,000 req/sHigh-throughput stock locks, Redis cache
Notification Service
24 GB5,00015,000 req/sAsync worker, Kafka consumer, Push/SMS
Search Service
416 GB1,5004,500 req/sOpenSearch query facade & indexing
Fleet Total29110 vCPU244 GB74,500 req/s1018% Headroom vs Peak