Given a stream of request/response samples, api-warden will:

• infer the request/response schema (json schema)
• infer the query string schema

This will be done per api-identifier. An api-identifier is constructed using this template: service|http-method|masked-uri|http-response-code

Everytime a change in schema is detected, a change event will be produced which will mark the change. It will include the before and after schemas.

kafka (traffic-samples) ---> api-warden ---> schema tracker
                                       |
                                        ---> kafka (schema change event)

Traffic sample schema

the incoming samples should follow this schema

• source (string): Identifier for the entity/component/service which you want to track the schema for
• uri (string): URI of the request. Api-warden will attemp to normalize the uri. Check the URI Normalization section for mode details.
• qs (string): Actual query string (params) in their normal format: p1=v1&p2=v2...
• req_hdr (map): Key-Value pairs of request header. Api-warden will ignore the values.
• method (string): HTTP method. It will be always converted to uppercase.
• req (string): Actual body of the request. Can be empty or null
• resp_hdr (map): Similar to req_hdr
• resp (string): Actual body of the response. Can be empty of null
• resp_code (integer): HTTP response code

Run The Benchmark

cargo bench --bench benchmark -- --profile-time=10