How Pub/Sub works

[vikhyat187]

can you assign this to me? @prakashchoudhary07

[vikhyat187]

[DRAFT]

• Pub / Sub stands for Publisher / Subscriber pattern.
• Its used to decouple services, make communication async.

Publishers communicate asynchronously with subscribers by producing and sending a message to a topic, which is a logical access point and communication channel. Subscribers (web servers, email addresses, Amazon SQS queues, Lambda functions) consume or receive the message or notification over one of the supported protocols (Amazon SQS, HTTP/S, email, SMS, AWS Lambda) when they are subscribed to the topic.

Sample code (AWS) for creating topics

Sample code SNS

• How can it be integrated with RDS repo
• Procedure to implement it
• How will the topic and response look like.
• In house/ managed pub sub service?

[vikhyat187]

Kafka vs Rabbit MQ [DRAFT]

While deciding between Apache Kafka vs RabbitMQ, it is to be noted that unlike Kafka, which can keep a large amount of data with minimum overhead, RabbitMQ queues are only quicker when they’re empty.

https://medium.com/analytics-vidhya/apache-kafka-architecture-getting-started-with-apache-kafka-771d69ac6cef https://hevodata.com/learn/kafka-vs-rabbitmq/ https://blog.logrocket.com/kafka-vs-rabbitmq-comparing-node-js-message-brokers/ https://www.notificationapi.com/blog/notification-service-design-with-architectural-diagrams https://tanzu.vmware.com/developer/blog/understanding-the-differences-between-rabbitmq-vs-kafka/ https://www.cloudamqp.com/blog/rabbitmq-use-cases-explaining-message-queues-and-when-to-use-them.html About Kafka https://www.confluent.io/blog/kafka-fastest-messaging-system/

---

Parameters	Kafka	RabbitMQ	SNS + SQS
Event microservices	supports	supports	supports
Stream processing	supports	supports (need to explore this)	not sure on this
Scale	Will be easily scalable	will be hard to scale	scalable as managed service
Retries	Will not support in house retries need to create explicitly	supports (need to explore on this)	supports dead letter queue
Fifo	Not sure	Not sure	Yes
Delivery latency	Less	not sure	Compared to kafka more
Cost	less	will be more on scale	compared to kafka more
Internal working	Stores in disk	stores In memory	not sure
mechanism	poll based	push based	not sure

[vikhyat187]

As per discussion with @prakashchoudhary07, we are going with Kafka.