Please enable JavaScript.

Coggle requires JavaScript to display documents.

MRTTLN DVA-C02, Exponential backoff - Coggle Diagram

- - - - Liner
      - AllAtOnce deploys to all instances at once
      - OneAtATime default configuration, one instance at a time
      - Canary
      - HalfAtATime deploys to half of the instances at a time, if odd number it will round up
    - - In-Place update/replace new app in same environment
      - Blue/Green
      - Cattle clound native architecture, replaceable units
      - Warm Standby scaled-down, ready-to-go environment in a secondary location
      - Pilot Light minimal version in a secondary region/ infrastructure
  - - - AppSync (GraphQL APIs)
      - HTTP/ REST
      - WebSocket
- - - - Pessmistic Locking
      - Idempotent Writes
      - Optimistic Concurrency Control natively support, multiple client attempt to update an item simultaneously
      - Batch Operations
- - - - Key Policies JSON policy control access to AWS KMS
  - - - RDS for Postgres
      - RDS for MySQL
- - - - Managed Node Groups creates & manages Nodes (EC2 Instance) for you
      - Self-Managed Nodes
      - AWS Fargate: no maintenance required; no nodes managed
  - - - Spread palce tasks evenly based on the specified value, key-value pair
      - Random
      - Binpack minimizes the number of instances and allow you to-be cost efficient, place tasks based on the least available amount of CPU/ memory
      - Task Palcement Constraint
        
        DistinctInstance
        
        MemberOf: deploy on instance that satisfy
- - - - Environment variable max 4KB
      - Max size /tmp 10GB
    - - Lambda Destination route the result of asynchronous function invocation to different AWS service
- - - - Delivery how message deliver from SNS to SQS, Retry Attempts, Error Handling
      - Redrive managing message processing failure, move to DLQ
    - - Long Polling its allow consumer to wait for message become available rather than returning immediately with an empty response if no message are present at the time of the request
      - SQS Extended Client Libary to send message large than 256KB by default
      - SQS KMS Encryption
    - - Attributes
        
        MessageGroupId ensure multiple messages belonging to the same group are processed in order
        
        MessageDeduplicationId
    - - PurgeQueue delete all message without deleting the queue itself
      - DeleteQueue delete the queue
- - - - CW alarm 10sec
  - - - Metadata also key-value pair but isnot indexed for seraching
      - Annotations key-value pair can associate with your trace, indexed by X-Ray. They can be searched & fitltered.
    - - Sampling is a process of selecting a subset of requests to be traced, helps manage costs & performance by limiting the amount of trace data sent to X-Ray
      - X-RAY SDK records 1st request each second and 5% of any additional requests
      - Segments each segment associated with a trace, such as an incoming request to a web service
- - - - Caching per stage, TTL 300s
- - - - volatile-LRU
      - allkeys-LFU
      - allkeys-LRU
      - volatile-LFU
      - volatile-TTL
      - noeviction
      - volative-random
      - allkeys-random
    - - Time based Expiration TTL
      - Manual Invalidation
      - Write-Through Caching
      - Cache Aside Pattern
      - Event-Driven Invalidation
    - - Write-Through Caching
      - Write-Back Caching
      - Lazy Loading (Cache Aside)
      - Read-Through Caching
      - Cache Expiration and TTL