Application Integration

Amazon EventBridge

  • Problems with monoliths

    • tightly coupled
    • single point of failure
    • scaling certain parts is not possible
  • Monolith ------> Microservices
  • Each Microservice would have their teams, resources etc.

  • Issues with Microservices

    • Noisy communication between several services might introduce inefficiencies.
    • Sometimes team communication becomes difficult in microservices.

Event Based Architecture

  • Services are more decoupled
  • Services will wait for a flag (event) to be sent from upstream and then action will be taken based on event.

Amazon EventBridge : An event co-ordinator

  • Events are json made of strings and define some activity
  • [IMP] : version, id, detail_type, src, account, time, region, resources, details : are common option among all events.
  • The only event-based service that integrates directly with third party support. Eg. Zendesk, Auth0, Segments, etc.

Event Bus : Like a funnel that catches events generated by all AWS Services.

  • Each account comes with a default event bus.
  • Event Bus has upto 100 event rules.

Rule : allows to filter a large number of events.

  • [IMP] Rules are not processed in an order.
  • We need to specify targets. (EC2, Kinesis, ECS, Lambda)

AWS Supports Cross-Account Events.

Event Bridge is just CloudWatch with new features and support for SAAS(Software as a Service) products.

Replaying Events and EventBridge Archives

  • Event Bridge Archives
    • Greate to store as a backup
    • No Cost to Store
    • Replay Events
  • Schema registry : Defines the structure of Event and helps describes what can you expect from event.
    • comes with pre-written schema for all AWS Service, and for few SAAS providers.
    • Schema can be generated automatically.

SNS (Simple Notification Service)

  • Publish/subscribe messaging service
  • managed, highly scalable

image-20230218165652816

  • SNS is mostly used as a producer for SQS Queue
  • Another Pattern
SNS ----> Lambda (modify notifications) ---->  SNS (new)
                                            |----> May take some action

SQS (Simple Queue Service)

  • capability of sending storing and recieving messages at scale without dropping message data.
  • Dead-letter-queue : Queue where dead letters/letters which are not cosumed are put at.
    • DLQ must be same type as the queue it is supposed to get letters from.

MODEL

  • Producers puts data in Queue
  • Consumers consumes data from Queue
    • Deletes data from queue : If processed
    • Puts back : If not processed

Standard Queue

  • At least once delievary of message
  • provides almost unlimited number of txn/sec
  • Best Effort to preserve order (not necessary)

FIFO Queue

  • Order is maintained (condition)
  • limited #TXN/sec (default 30 TPS)
  • Batching allows to perform action against 10 messages at once with a single action
  • Exactly once processing

AWS Step Functions

  • Lambda has limited flexibility and execution time limit of 15 min.
  • Step Function : State Machine Services
    • Parallel
    • Sequential
    • In-Retry
    • If-Then
  • States of Step Functions :
    1. PASS
    2. TASK
    3. CHOICE
    4. WAIT
    5. SUCCESS
    6. FAIL
    7. PARALLEL
    8. MAP

Important Comparisons for AWS Exam

EventBridge

  • A very complex and sophisticated service

SNS

  • limited parameters
  • easily scales to millions
  • No direct connectivity to Saas
  • Limited Routing Capability
  • Difficult to trigger Step Function
  • Filtering is limited to attributes only