10. Performance and Scalability
The Hopex MCP Server runs in HAS. Performance depends on the selected MCP tool, GraphQL query, returned volume, relationship depth, schema, request precision, and Hopex-side processing cost.
10.1. Web API-equivalent behavior
An MCP call is processed by the MCP Server and relayed to Hopex through GraphQL/Web API or exposed Hopex services. This execution chain uses the same technical framework as existing Web API calls.
MCP load mainly depends on the Hopex operation being executed. A simple query on a limited dataset is comparable to a simple Web API call. A broad query, deep relationship traversal, or complex mutation generates higher load.
10.2. Performance factors
The main performance factors are the tool being called, the GraphQL schema used, the applied filters, the number of returned results, the relationship depth, and the processing cost on the Hopex side.
Introspection and guide tools mostly return schema metadata. GraphQL execution tools depend directly on the submitted query. Diagram tools depend on the target objects, associated diagrams, and requested output format.
10.3. Data volume management
The MCP Server uses GraphQL filters, pagination, and result limits to reduce returned volume and server load.
Broad user requests may trigger expensive calls. Tool descriptions, query guides, and schema metadata help guide the AI client toward precise requests.
10.4. Load, concurrency, and supervision
MCP calls consume the same platform resources as Web API access: HAS execution, repository access, GraphQL/Web API processing, response generation, and logging. Concurrency is analyzed through the standard HAS and Hopex load model.
HAS supervision tracks MCP sessions, clients, tools, resources, execution results, and estimated volumes. These events help relate MCP usage to Hopex service load.
10.5. Write-operation impact
When mutations are authorized, the performance impact follows the same logic as write operations performed through the Web API. The cost depends on the modified object type, application checks, impacted relationships, and repository traceability mechanisms.
Mutation enablement is defined in section 5.1.2 Mutation authorization and governed in section 13.4 Write-operation governance.
The mutation tool call is traced on the MCP side. The actual changes are tracked by Hopex mechanisms at the level of the affected objects, as with other modifications performed through Hopex services.