AWS Neptune Sizing Calculator

AWS Neptune Sizing Calculator MCP Connector for Claude

A+

Deterministic sizing for AWS Neptune graph databases.

3 tools Official Updated Oct 1, 2026 Official Vinkius Partner

This MCP server provides deterministic sizing and configuration estimates for AWS Neptune graph database deployments. Use calculate_storage_needs to determine the storage footprint and memory requirements based on vertex and edge counts. Use estimate_cluster_configuration to get recommended replica counts and snapshot retention settings. You can also use evaluate_performance_benchmarks to find optimal thresholds for bulk loading and query sizes. It helps ensure your Neptune cluster is correctly provisioned for both storage and high availability.

neptunegraphawssizingdatabase

3 tools expose this connector's capabilities to your AI agent.

calculate_storage_needs

Determines the total storage footprint and the necessary memory overhead for the database

estimate_cluster_configuration

Recommends the instance scaling, replication, and operational settings for the cluster

evaluate_performance_benchmarks

Provides reference values for workload performance and query sizing

See how to talk to your AI agent using AWS Neptune Sizing Calculator.

Calculate the storage and memory needs for a graph with 1,000,000 vertices and 5,000,000 edges, with 5 properties per vertex and 2 per edge, at 50 concurrent queries.

The total storage estimation is 150,000,000 bytes and the required memory is 300,000,000 bytes.

What are the recommended performance benchmarks for a Neptune workload?

Recommended benchmarks include a concurrency of 1,000, a bulk loader throughput of 100,000 triples per second, and a 1 MB limit for both Gremlin and SPARQL queries.

Recommend a cluster configuration for a high availability deployment with 7 days of snapshot retention.

For high availability, the recommended replica count is 1 and the snapshot retention is confirmed for 7 days.

Storage is calculated by summing the total size of all vertices and edges, where each unit is weighted by its property count.

Related Connectors