ShoalStack is now
ShoalStack is now
A message queue keeps slow work from delaying an API or another service, but it is one more cloud resource to build. platform201 builds it for you: add the queue to the visual graph, and then deploy.
A producer service sends messages, and a consumer service reads them. Place a queue between them on the graph, and platform201 creates it when you deploy, so you do not build it by hand.
Add a queue node, and set how long it keeps messages. The node stays on the graph next to your services, so you can see what connects to it.
Connect the queue to each service. The connection passes the queue values to the variable names your code reads, so you do not copy them yourself.
A queue has no direct connection to a database. Connect the database to the consumer, and the consumer saves the data your application keeps.
Deploy the queue, your producer, your consumer, and your database as one environment. Then follow each deployment in its status and its log.
A queue is most useful when one service should not wait for another service. The producer sends a message and continues, and the consumer reads the message when it is ready.
An API sends a message and responds at once. A container or a function does the slow job as the worker.
A spike of requests waits in the queue. A worker reads messages at its own pace and saves the results.
A Node.js API and a Python or Go worker share one queue. The services agree on the message format.
If your API accepts image jobs, a worker can save each result to Cloud SQL. A queue connects the two on the graph, and the API keeps responding while the worker is paused.
A new environment has a gateway connected to a container. Use that container as the API. Then add a worker, a queue, and a Cloud SQL database.
Connect the queue to the API. Map the queue values to the variable names your API code reads. The API code sends a message for each image job.
Add a subscription to the queue. The worker reads its messages there. Connect the queue and the database to the worker, and map their values to variables.
Deploy the environment. Pause or redeploy the worker. Keep sending image jobs. The API keeps responding, and the worker saves the results when it returns.
A queue is one node on the graph. The gateways, services, and databases around the queue form your application. You deploy them in one environment and follow each deployment.
Add a queue node with one subscription, and connect a producer to it. Then connect a consumer to read the messages. Connect a database to the consumer when you need one.
A background worker is a service that does slow jobs outside a web request. An API accepts the request, responds at once, and leaves the job for the worker. Common jobs include resizing an image or creating a report, because a user should not wait for them. In platform201, a worker is a container or a function that you connect to a queue on the graph.
