Request Per Second Calculator

A database and container sizing calculator estimates the key capacity numbers for distributed data infrastructure: the number of database shards required for a given dataset, B-tree index storage size, request throughput (RPS) for a given concurrency model, CPU and memory totals for Docker or Kubernetes deployments, expected replication lag under a given write rate, and the failure rate that will trip a circuit breaker. These calculations underpin capacity planning for microservices, cloud-native databases, and container orchestration platforms.

Related calculators: the API and cloud cost calculator for estimating the monthly cost of the infrastructure being sized, and the DevOps metrics calculator for tracking operational health after deployment.

1. Select the tab for your sizing problem: Shard Count, Index Size, RPS Budget, Container Resources, Replication Lag, or Circuit Breaker.
2. For shard count, enter your total dataset size and the target maximum shard size — results include replication overhead.
3. For RPS, enter peak concurrent users, requests per session, and session duration — the calculator estimates required server throughput.
4. For container resources, enter pod count and per-pod CPU/memory requests to get cluster-level totals for node sizing.

Database and container sizing formulas

Shard count — shards = ⌈total_data_GB / max_shard_GB⌉; total with replicas = shards × replication_factor

Index size (B-tree) — index_bytes = rows × avg_key_bytes × overhead_factor (typical 2–3×)

RPS throughput — RPS = concurrent_users × requests_per_session / session_duration_seconds

Container totals — total_CPU = pods × cpu_per_pod (cores); total_RAM = pods × ram_per_pod (MiB)

Replication lag — lag_ms ≈ (write_MB_s / link_MB_s) × 1000 + RTT_ms

Circuit breaker — failure_rate = failures / total_requests × 100%; trips to Open when rate ≥ threshold

Interpreting sizing results

Sharding and throughput guidelines

For Elasticsearch, keep individual shards between 10 and 50 GB and limit total shard count to approximately 20 per GB of JVM heap. For Postgres with partitioning or Citus sharding, 25–100 GB per shard is typical depending on query pattern. An RPS figure from the throughput tab tells you what your application servers must sustain at peak — compare it against your framework's benchmarked single-instance throughput to determine how many application instances (pods) you need. For circuit breakers, a 50% failure rate threshold is a common default (Hystrix, Resilience4j), but latency-sensitive services often use a lower threshold of 20–30% to fail fast and protect downstream systems.

Technology tips and best practices

• When sizing Kubernetes pods, set resource requests at 50–70% of expected steady-state usage so the scheduler can place pods effectively. Set limits at 1.5–2× requests to absorb traffic bursts without OOM kills.
• Use the cloud cost calculator alongside this tool: once you know the pod count and node sizes needed, you can estimate the monthly cloud bill for the deployment.
• Replication lag spikes during batch inserts or schema migrations. If link saturation (write rate / bandwidth) exceeds 70%, consider throttling batch operations to off-peak hours or splitting large migrations into smaller transactions.
• Circuit breaker half-open state (testing if the downstream service has recovered) should use a small number of probe requests — typically 1–5 — before deciding whether to fully close the circuit. Configure probe timeout shorter than normal to avoid cascade delays.

Database and infrastructure sizing figures are estimates based on simplified models. Real-world performance depends on query patterns, hardware I/O characteristics, network topology, and software versions. Validate estimates with load testing and monitoring before committing to production capacity decisions.

Related Calculators

• Download Time Calculator
• Network Bandwidth Calculator
• Storage Calculator
• Battery Life Calculator