Kafka-Message-Server | Example application based on Apache Kafka framework | Pub Sub library

Because your pipeline does not have any unbounded PCollections, it will be automatically run in batch mode. You can force a pipeline to run in streaming mode with the --streaming command line flag.

If you check the bottom of your Google Cloud console screenshot, you can see it suggests to have 4 partitions with 4MiB/s publish and subscribe throughput.

Therefore your Terraform partition_config should match this. Count should be 4 for the 4 partitions, with capacity of 4MiB/s publish and 4MiB/s subscribe for each partition.

The "peak throughput" in web UI is just for convenience to help you choose some numbers here. The actual underlying PubSub Lite API doesn't actually have this field, which is why there is no Terraform setting either. You will notice the sample docs require a per-partiton setting just like Terraform.

eg. https://cloud.google.com/pubsub/lite/docs/samples/pubsublite-create-topic

I think the only other alternative would be to create a reservation attached to your topic with enough throughput units for desired capacity. And then completely omit capacity block in Terraform and let the reservation decide.

Should be:

nsc edit user RequestCacheService --allow-pubsub '$JS.API.STREAM.INFO.*'

With single-quotes around the subject. I was under the impression that double & single quotes would escape the $ but apparently only single-quote will escape special characters in the subject.

MSK VS SQS+SNS is not really 1:1 comparison. The choice depends on various use cases. Please find out some of specific difference between two

1. Scalability -> MSK has better scalability option because of inherent design of partitions that allow parallelism and ordering of message. SNS has limitation of 300 publish/Second, to achieve same performance as MSK, there need to have higher number of SNS topic for same purpose.

Example : Topic: Order Service in MSK -> one topic+ 10 Partitions SNS -> 10 topics

if client/message producer use 10 SNS topic for same purpose, then client needs to have information of all 10 SNS topic and distribution of message. In MSK, it's pretty straightforward, key needs to send in message and kafka will allocate the partition based on Key value.

1. Administration/Operation -> SNS+SQS setup is much simpler compare to MSK. Operational challenge is much more with MSK( even this is managed service). MSK needs more in depth skills to use optimally.

2. SNS +SQS VS SQS -> I believe you have multiple subscription(fanout) for same message thats why you have refer SNS +SQS. If you have One Subscription for one message, then only SQS is also sufficient.

3. Replay of message -> MSK can be use for replaying the already processed message. It will be tricky for SQS, though can be achieve by having duplicate queue so that can be use for replay.

It seems that the numbers you mentioned, correspond to not customizing the job resources. By default streaming jobs use a n1-standar-4 machine:

3 Streaming worker defaults: 4 vCPU, 15 GB memory, 400 GB Persistent Disk.
4 vCPU x 24 hrs x 21 days = 2,016
15 GB x 24 hrs x 21 days = 7,560

If you really need streaming in Dataflow, you will need to pay for resources allocated even if there is nothing to process.

Options:

Optimizing Dataflow

• Considering that the number and size of the JSON string you need to process are really small, you can reduce the cost to aprox 1/4 of current charge. You just need to set the job to use a n1-standard-1 machine, which has 1vCPU and 3.75GB memory. Just be careful with max nodes, unless you are planning increase the load, one node may be enough.

Your own way

• If you don't really need streaming (not likely), you can just create a function that pulls using Synchronous Pull, and add the part that writes to BigQuery. You can schedule according to your needs.

Cloud functions (my recommendation)

• You can create a serverless Event-Driven Cloud Function with a Cloud Pub/Sub trigger. This way, considering your low volume, you can take advantage of the Free Tier and keep the real time processing:

"Cloud Functions provides a perpetual free tier for compute-time resources, which includes an allocation of both GB-seconds and GHz-seconds. In addition to the 2 million invocations, the free tier provides 400,000 GB-seconds, 200,000 GHz-seconds of compute time and 5GB of Internet egress traffic per month."[1]

[1] https://cloud.google.com/functions/pricing

I eventually went for implementing a Kafka connector myself and using Kafka.

This is not possible.

You cannot have a single registration to track course work changes for multiple courses, as you can see here:

Types of feeds

The Classroom API currently offers three types of feed:

• Each domain has a roster changes for domain feed, which exposes notifications when students and teachers join and leave courses in that domain.
• Each course has a roster changes for course feed, which exposes notifications when students and teachers join and leave courses in that course.
• Each course has a course work changes for course feed, which exposes notifications when any course work or student submission objects are created or modified in that course.

If you think this feature could be useful, I'd suggest you to file a feature request in Issue Tracker using this template.

• Types of feeds
• Missing features

You have a Flow> here, which means every element of this flow is itself a list.

The take operator is applied on the flow, so you will take the 3 first lists of the flow. Each individual list is not limited, unless you use take on the list itself.

So the name transformedListOf3Elements is incorrect, because the list is of an unknown number of elements, unless you filter it somehow in the map.

To feed a Source, you don't necessarily need to use a custom graph stage. Source.queue will materialize as a buffered queue to which you can add elements which will then propagate through the stream.

There are a couple of tricky things to be aware of. The first is that there's some subtlety around materializing the Source.queue so you can set up the subscription. Something like this: