Exam Professional Cloud Architect topic 4 question 5 discussion

Answer : C (https://cloud.google.com/pubsub/docs/publisher?hl=en#batching) Cost of Batching is latency for individual messages,. To minimize latency batching should be turned off

C - The cost of batching is latency for individual messages, which are queued in memory until their corresponding batch is filled and ready to be sent over the network. To minimize latency, batching should be turned off. https://cloud.google.com/pubsub/docs/publisher?hl=en#batching A incorrect. Application timeout because of publisher latency, nothing to do with timeout retry with publish request. D does not make sense at all. B is about receiver, not publisher.

The issue at hand is increased load causing timeout errors during interactions with Pub/Sub. The lack of Pub/Sub publishing errors indicates that the problem is likely related to retries or network performance. By increasing the Total Timeout retry value, you allow the application more time to complete message publishing during high-load scenarios, reducing the chances of timeouts

"the application is not loggin any Pub/Sub publishing errors", so no need to increase Pub/Sub Total Timeout.

Latency in Pub/Sub can be of two types: End-to-end latency is the time it takes for a message to be published by a publisher and delivered to the corresponding subscribers for processing. Publish latency is the amount of time it takes to publish a message. When using batching, increasing both types of latencies is a trade off for improving efficiency and throughput.

C https://cloud.google.com/pubsub/docs/publish-best-practices#configure-batch

See Solution #1 below link https://saturncloud.io/blog/how-to-fix-deadlineexceeded-when-publishing-to-a-cloud-pubsub-topic-from-compute-engine/#:~:text=The%20timeout%20limit%20for%20publishing,receive%20the%20DEADLINE_EXCEEDED%20error%20message.

See video "Cloud Pub/Sub Publishers - ep. 4" at 2m 25 sec where Priyanka says "to reduce latency, batching should be turned off" https://www.youtube.com/watch?v=ML6P1ksHcqo&list=PLIivdWyY5sqKwVLe4BLJ-vlh9r9zCdOse&index=4. Batching increases throughput but adds latency for individual messages as they are queued in memory until their batch is filled.

I voted C but I think it has to be A. "Total timeout: the amount of time after a client library stops retrying publish requests." "After each publish request, the request timeout increases by the request timeout multiplier, up to the maximum request timeout." So it's increasing the timeout value on the retry which seems like the best solution.

B. Move from a Pub/Sub subscriber pull model to a push model. Explanation: Moving from a pull model to a push model in Google Cloud Pub/Sub can help improve the latency in your application. In a push model, the messages are pushed from the Pub/Sub service to the subscriber application, reducing the time it takes for the application to receive messages. This can help mitigate the timeout errors that you are experiencing due to increased load on the application servers.

This is chatgpt's choice I agree that Pub/Sub message batching can be a useful optimization for improving overall throughput and reducing the number of API calls required to publish messages. However, in the context of addressing timeout errors during publishing, turning off message batching may not be the most appropriate solution. In cases where message batching is causing issues, such as network or system resource constraints, reducing the batch size or adjusting the batch duration can help improve publishing latency. However, in the case of timeout errors, increasing the Total Timeout retry value would be a more effective solution, as it allows more time for the message to be successfully published and reduces the likelihood of encountering timeout errors.

To improve publishing latency in this scenario, it is recommended to turn off Pub/Sub message batching. By turning off message batching, you can send messages individually as soon as they are published, rather than waiting for a batch of messages to be created before sending them. This can help to reduce the risk of timeout errors and improve the overall performance of the application. It is also a good idea to monitor the application's performance and error logs to identify any other potential issues that may be contributing to the timeout errors.

C is the correct answer

but how is this related to turnoff batching??

C is ok

But no where in the question or scenario it says they turned on Pub/Sub batching.

C is better option, even though increasing total timeout would help reduce timeout errors but remember that that in this case we are getting too many messages from the server since load increased and we need to reduce latency