Sharing of flows in a stream processing system

摘要

Techniques for generating code for a flow are provided. The techniques include receiving a flow for an application to be deployed, wherein the flow comprises one or more sub-flows, identifying sub-flows of the flow that match existing sub-flows that are already running in separate flows by querying a run-time flow information repository, generating code for the flow, wherein the code comprises newly generated code for each of the sub-flows that do not match existing sub-flows that are already running in separate flows, and wherein the code also comprises one or more connectors to matching sub-flows that are already running in separate flows in lieu of generating new code for the sub-flows that match existing sub-flows that are already running in separate flows, and deploying the generated code for the flow and updating the run-time flow information repository with information pertaining to the flow.

主权项

1. A method for generating code for a flow, wherein the method comprises:
receiving a flow for an application to be deployed, wherein the flow comprises one or more sub-flows; identifying one or more of the sub-flows of the flow that match a respective one of one or more existing sub-flows that are already running in one or more separate flows by querying a run-time flow information repository; generating code for the flow by (i) generating new code for each respective one of the one or more sub-flows that do not match a respective one of the one or more existing sub-flows that are already running in the one or more separate flows, (ii) generating one or more import/export statements that establish a connection to each respective one of the one or more existing sub-flows that are already running in the one or more separate flows that match a respective one of the, and (iii) inserting a respective one of the one or more generated import/export statements in each of one or more positions in the code that correspond to each respective one of the one or more sub-flows that match a respective one of the one or more existing sub-flows that are already running in the one or more separate flows; annotating one or more components of the generated code that will be active in the run-time for more than a predetermined period of time, wherein said annotating comprises applying an @share annotation to each of the one or more components; specifying one of multiple sharing contexts to be attributed to each of said annotated components; and deploying the generated code for the flow and updating the run-time flow information repository with information pertaining to the flow.