Gateway service driven downstream orchestration
Various techniques are disclosed for providing gateway services between a client system and downstream service systems for a service system. The disclosed gateway service system is capable of implementing a new service for a client system in response to a request from the client system. The gateway service system internally determines which downstream services are needed to implement the new service. In various instances, the gateway service system utilizes machine learning algorithms to determine the downstream services suitable for providing the output needed for the new service. The gateway service system is also capable of determining whether an existing application programming interface (API) is able to be used for the new service or whether a new API needs to be created for the new service. By internally determining the downstream services and APIs, the gateway service system has more efficient utilization of its computational resources.
1 . A method, comprising:
receiving, by a computer system of a service system, a request from a client system to process particular service requests from the client system at the service system in association with downstream services, wherein the request includes information on one or more client system characteristics associated with the particular service requests;
determining, at the computer system, an expected output object for an application programming interface (API) to handle the particular service requests based on the information on the one or more client system characteristics;
accessing, at the computer system, information corresponding to one or more of the downstream services, the information including output schema for the downstream services;
determining, at the computer system and based on the output schema of the downstream services, at least one set of downstream services capable of providing the expected output object for the API, wherein the at least one set of downstream services has a confidence score that satisfies a predetermined threshold, the confidence score being an indication of a predicted confidence that the at least one set of downstream services will provide the expected output object for the API;
selecting, at the computer system, a specified set of downstream services to provide the expected output object for the API from the at least one set of downstream services by applying one or more particular rules; and
determining, at the computer system, the API to implement for the client system based on the specified set of downstream services, wherein determining the API to implement includes one of:
identifying an existing API capable of providing the expected output object using the specified set of downstream services; and
generating a new API capable of providing the expected output object using the specified set of downstream services.
2 . The method of claim 1 , wherein determining the at least one set of downstream services capable of providing the expected output object for the API includes:
applying the expected output object for the API and the output schema of the downstream services as inputs to a machine learning algorithm;
obtaining outputs of one or more sets of downstream services capable of providing the expected output object for the API; and
selecting as the at least one set of downstream services, one of the one or more sets of downstream services with its confidence score exceeding the predetermined threshold.
3 . The method of claim 1 , wherein the at least one set of downstream services includes a single downstream service or a combination of multiple downstream services.
4 . The method of claim 1 , wherein determining the expected output object for the API includes:
applying one or more processing algorithms to the one or more client system characteristics to enrich the system characteristics, at least one processing algorithm being a text processing algorithm; and
implementing a machine learning algorithm on the enriched system characteristics to determine the expected output object for the API.
5 . The method of claim 1 , wherein the one or more client system characteristics include one or more of the following characteristics: sample data available for the particular service requests, service level agreement data for expected responses from the API, expected volume of requests to the API, and an intent for the particular service requests.
6 . The method of claim 1 , wherein applying the one or more particular rules includes filtering the at least one set of downstream services according to criteria in the particular rules.
7 . The method of claim 6 , wherein the criteria in the particular rules includes one or more of the following criteria: latency threshold, privacy threshold, and service level agreement data thresholds.
8 . The method of claim 1 , wherein the generation of a new API for the client system is implemented when the specified set of downstream services includes multiple downstream services.
9 . The method of claim 1 , wherein the existing API is associated with a single downstream service.
10 . The method of claim 1 , wherein the API is operable to:
receive service requests from the client system;
transmit requests for output responses to the downstream services in the specified set of downstream services;
receive the output responses from the specified set of downstream services; and
transmit the received output responses to the client system.
11 . A gateway service system, comprising:
a non-transitory memory; and
a processor coupled to the non-transitory memory and configured to read instructions from the non-transitory memory to cause the gateway service system to:
receive a request from a client system to process particular service requests from the client system in association with downstream services of the gateway service system, wherein the request includes information on one or more client system characteristics associated with the particular service requests;
apply at least one first machine learning algorithm to the client system characteristics to determine an expected output object for an application programming interface (API) to handle the particular service requests;
access information corresponding to one or more of the downstream services, the information including output schema for the downstream services;
apply the expected output object for the API and the output schema of the downstream services as inputs to at least one second machine learning algorithm;
select, based on outputs from the at least one second machine learning algorithm, at least one set of downstream services for providing the expected output object for the API, wherein the outputs from the at least one second machine learning algorithm include combinations of one or more downstream services capable of provided the expected output object for the API along with confidence scores for the combinations, the confidence scores being indications of predicted confidences that the combinations of one or more downstream services will provide the expected output object for the API; and
apply one or more predetermined rules to the at least one set of downstream services to determine a specified set of downstream services to provide the expected output object for the API from the at least one set of downstream services;
determining a suggested implementation of the API for the client system based on the specified set of downstream services, wherein determining the API to implement includes;
identifying an existing API that is capable of providing the expected output object using the specified set of downstream services; and
when the existing API cannot be identified, generating a new API capable of providing the expected output object using the specified set of downstream services;
generating a report on the suggested implementation of the API for the client system; and
transmitting the report to the client system for presentation to a user of the client system.
12 . The gateway service system of claim 11 , wherein the gateway service system selects the at least one set of downstream services based on the confidence score exceeding a predetermined threshold.
13 . The gateway service system of claim 11 , wherein the predetermined rules include rules that filter the at least one set of downstream services according to criteria.
14 . The gateway service system of claim 11 , further comprising instructions that cause the gateway service system to receive, from the client system, a response indicating to proceed with implementation of the API in the gateway service system.
15 . The gateway service system of claim 11 , wherein the report is presented on a user interface of the client system.
16 . The gateway service system of claim 11 , wherein the suggested implementation of the API in the report includes a recommendation to modify one or more downstream services, or a recommendation for new downstream services.
17 . A non-transitory machine-readable medium having stored thereon machine-readable instructions executable to cause a machine to perform operations comprising:
receiving a request from a client system to process particular service requests from the client system in association with downstream services for the service system, wherein the request includes information on one or more client system characteristics associated with the particular service requests;
determining, by implementing a first machine learning algorithm on the information on the one or more client system characteristics, an expected output object for an application programming interface (API) to handle the particular service requests;
accessing information corresponding to one or more of the downstream services, the information including output schema for the downstream services;
determining, by implementing a second machine learning algorithm on the expected output object for the API and the output schema of the downstream services, at least one set of downstream services capable of providing the expected output object for the API, wherein the at least one set of downstream services has a confidence score that satisfies a predetermined threshold, the confidence score being an indication of a predicted confidence that the at least one set of downstream services will provide the expected output object for the API;
selecting a specified set of downstream services to provide the expected output object for the API from the at least one set of downstream services by applying one or more predetermined rules; and
determining the API to implement for the client system based on the specified set of downstream services, wherein determining the API to implement includes either:
identification of an existing API capable of providing the expected output object using the specified set of downstream services; or
generation of a new API capable of providing the expected output object using the specified set of downstream services when an existing API is not identified.
18 . The non-transitory machine-readable medium of claim 17 , wherein the machine to perform operations is a gateway service machine for a service system.
19 . The non-transitory machine-readable medium of claim 17 , wherein the instructions include instructions for the machine to notify the client system of the selection of the specified set of downstream services.
20 . The non-transitory machine-readable medium of claim 17 , wherein implementing the API for the client system based on the specified set of downstream services include instructions for the machine to:
when the confidence score of the specified set in providing the expected output object for the API is above a second predetermined threshold, implement the downstream services in the specified set without modification; or
when the confidence score of the specified set in providing the expected output object for the API is below the second predetermined threshold, modify at least one of the downstream services in the specified set; and
implement the specified set with the modified at least one downstream service.