Ideal GCP infra for a microservice for invoking potentially long running tasks

[anonymous]

Hi everyone,

We’re looking to implement a new microservice in GCP to handle potentially long running tasks (anywhere from a few seconds to several days). This is our high level requirements, so far:

• [A] Able accept requests/tasks (not necessarily HTTP, “fire and forget”, to run our Python code for [B], [C], and [D]
• [B] Able to open and listen to a WebSocket connection from a third party API
• [C] Able to make HTTP calls to a third party API
• [D] Support terminating specific processes that have started (some other external service is able to target/terminate a running task)

Based on this information, what would be the ideal GCP infra for this new service?

We’ve narrowed down to a couple ideas and have some direction of what we believe will make a suitable MVP, but we’d also love to seek advice from the experts in this community as well.

Thank you and appreciate any advice!

[robertcarlos]

Hi @ian-ridian ,

Welcome to Google Cloud Community!

Here are some of the suggestions I could recommend for your new microservices:

• For [A]: I would suggest using Cloud Run as it supports [B], [C], and [D]. You can check this Cloud Run quickstart guide for Python for your reference.
• For [B]: you can check out this documentation on triggering websockets for Cloud Run if that would be applicable for your project.
• For [C]: you can check this Cloud Run Admin API for your reference.
• For [D]: you can check this documentation on managing services as it has the option to disable an existing service.

Let me know if you find the aforementioned steps helpful.