What it does
Platform administrators can post a workload description to /admin/v1/deployment-fit/evaluate and get a deterministic, scored comparison of deployment modes and of paths for bringing data in or mirroring it out. The score is weighted coverage of the needs you declare. It is not a benchmark. Hard disqualifiers, for example a requirement for the PostgreSQL wire protocol, are reported as such.
How it works
The catalogue is built from what the repository registers: the native deployment modes, the built-in inlets, the outbound sinks and the cluster admission rules. GET /admin/v1/deployment-fit/catalog returns the candidates, a classification of well-known systems (PostgreSQL, SQLite, MongoDB, MySQL, MariaDB and others, each marked with the inlets and mirrors that exist for it), the requirement dimensions, the workload presets, and the scoring rule.
POST /evaluate takes a request with the schema, catalogue and scorer versions it was written against, a mission (deploy_native, bring_data, keep_in_sync, mirror_out, sql_coexistence, edge_delivery or compare_only), an optional preset, a provider and a monthly budget, and a rating from ignore to required for any of eleven dimensions: local and offline use, durable restart, read scale, SQL wire protocol, BM25, vector search, auth, files, realtime, edge and operational simplicity. Unknown fields and unknown versions are rejected.
Each candidate gets full, partial, none or unknown support per dimension. The fit is the weighted sum of support divided by the weighted sum of the ratings. A required dimension that is not fully supported makes the candidate ineligible: it may show a score, but it never wins. The same input produces the same bytes, because the scorer uses no clock and no randomness, and golden tests pin that.
The advisor admits nothing. A client-supplied score is ignored, and the cluster admission rules alone authorize a deployment. Evidence that is performance-shaped is marked estimated or unknown, because no benchmark artifact is attached to a candidate.
Where you reach it
- HTTP
GET /admin/v1/deployment-fit/catalog - HTTP
POST /admin/v1/deployment-fit/evaluate
Example
{
"schema": "kynetra.deployment-fit/v1",
"catalog_version": "<value from GET /catalog>",
"scorer_version": "<value from GET /catalog>",
"mission": "deploy_native",
"requirements": {
"durable_restart": "required",
"vector_search": "high",
"sql_wire": "ignore"
}
}Limits
- A fit score is not a performance measurement.
- The catalogue is a snapshot compiled into the server and carries its own version string. It reflects what the repository registers at this commit.
- The routes need platform administrator authority.
Evidence
- On this siteDeployment target matrix
- TestsCovered in the engine repository by 22 tests (golden fixtures for determinism, unsupported systems getting no winning score, request validation) and a mounted-router test that the production app serves the routes with production authentication
Test coverage is listed as present in the engine repository at this edition. This page does not report test results.
KynetraDB is in private preview
Request a preview to try this against your own workload.