Abstract: A method, including receiving, from a client, a unified query, and extracting, from the unified query, an endpoint query for a first data source on a first server and an endpoint query for a second data source on a second server. The extracted endpoint query for the first data source is forwarded to the first server. Upon receiving a response to the endpoint query forwarded to the first server, one or more parameters are extracted from the response. The endpoint query for the second data source is updated so as to include the extracted one or more parameters, and the updated endpoint query for the second data source is forwarded to the second server. Upon receiving, from the second server, a response to the forwarded endpoint query, a result for the received unified query is generated based on the receive responses, and the generated result is conveyed to the client.

Abstract: A method, including receiving, from a client, a unified query, and extracting, from the unified query, an endpoint query for a first data source on a first server and an endpoint query for a second data source on a second server. The extracted endpoint query for the first data source is forwarded to the first server. Upon receiving a response to the endpoint query forwarded to the first server, one or more parameters are extracted from the response. The endpoint query for the second data source is updated so as to include the extracted one or more parameters, and the updated endpoint query for the second data source is forwarded to the second server. Upon receiving, from the second server, a response to the forwarded endpoint query, a result for the received unified query is generated based on the receive responses, and the generated result is conveyed to the client.

Abstract: A method, including receiving, by a first API hub server in communication with a set of private APIs and with a second API hub server that is in communication with a set of public APIs, metadata for the APIs, each of the APIs including one or more respective API calls. A request is received from a client computer to access a given endpoint of a given API, and using the metadata for the given API, an API call is generated. Upon determining that the given API includes a given private API the generated API call is conveyed to the given private API, and upon determining that the given API call includes a given public API, the generated API call is conveyed to the second API hub server. Finally, upon receiving a response to the conveyed API call, the response is forwarded to the client computer.

Abstract: A method for ranking a set of APIs for multiple software applications executing on respective servers, including initializing, by a first computer, respective ranking scores for the APIs, and receiving, from a plurality of second computers, multiple API calls to the APIs in the set. For each given API call to a given API in the set received from a given second computer, a given server executing a given software application including the given API is identified, the given API call is forwarded to the identified server, a response to the given API call is received from the identified server, performance metrics with respect to the given API call and response are computed, and the ranking score for the given API are updated based on the computed performance metrics. The APIs can be provided to the second computers responsively to the ranking scores, as updated in response to the requests.