Applications Flashcards
Describe SQS and their queues and features?
It is the queueing system AWS offers. It is used to decouple components. In a nutshell, a producer puts the message which is durable. One or many readers keep asking for messages.
SQS is pull-based. It means you need an application to ask for a message. It is not pushed-based as SNS.
The messages are 256KB in size, but if needed, an increase can be requested
The TTL is 1min to 14 days. The default retention period is 4 days
Visibility timeout: By default, it is 12 hours. It is the time the message is invisible before timeout. Otherwise, in case the message is not processed, it will be available for another reader
The SQS long pooling can be used to reduce the billing. It does not return until a message arrives or the long poll times out
There are two types of queues:
Standard queue: unlimited number of transactions per second. But there is a possibility of duplications and out of order messages
FIFO: It is limited to 300 transactions per second. The message remains available until a consumer processes and deletes it. There is no duplication
What Simple Work Flow Service (SWS) is and what are the actors?
It is a task-oriented system that coordinates the execution of tasks (sync/async). It also supports human-task interaction.
It has the following actors:
Workflow starters: An application that initiates a workflow;
Deciders: It decides what to do in case of something finishes/fails;
Activity workers: It processes/carry out the activity tasks;
An workflow execution can last up to 1 year.
Before creating a workflow, one needs to register a domain on which you will define the retention period
Simple Notification Service
It supports push notifications to mobile devices, SMS, email, SQS queues, or any HTTP endpoint
Elastic Transcoder
It is a workflow that allows format conversion with low cost. A simple use case it to record in 4K and converts to optimal formats depending on the device type (smartphones, tables, PCs, etc).
Why you’d use AWS API Gateway?
To scale and expose HTTPS RESTful API which usage is tracked, throttled, logged via API key. It maintain multiple versions of API for staging/production
You can configure one by defining the API, the (nested) resources and their HTTP methods, security, target, and request/response transformation
It provides free SSL/TLS in case it is used with AWS Certificate Manager
It supports cache for endpoints responses
What CORS?
It is the mechanism to allow restricted resources on a website to be requested from another domain outside the domain from which the first resources was served.
How it works?
1: The browser makes an HTTP Options call for an URL
2: The responds with “These other domains are approved to GET this URL”
3: Error-“Origin policy can’t be read at the remote resource”
Describe Kinesis
Streaming data is generated by thousand of data sources such as stock prices, game data (as the gamer plays), social network, geospatial, IoT sensor data, etc
Kineses has 1-7 days retention.
There are three types of Kinesis:
Streams: The producers put the data into shards (topics). The consumers get the shards data to process it. 1000 writes/second. 5 reads/second and 2 MB/second and outputs to DynamoDB/EMR/S3/Redshift
Firehose: Similar to Streams, but has no persistence/shards. The data needs to processed/converted right away (with lambda) and outputs to S3/Elasticseach
Analytics: It analyses (SQL or Flink) the data on the fly and outputs to S3/Redshift/Elasticseach
Describe Cognito
Web Identify Federation that is recommended for all mobile applications.
There are flavours:
User pools: Sign-in/up functionality
Identity pools: The user enters his credentials into the social media (Web Identity Provider), Cognito receives an authentication code which is exchanged for AWS credentials allowing them (users) to assume an IAM role.
What is the size limit an item in DynamoDB?
DynamoDB supports single-digit millisecond latency at any scale because it is stored in SSD across three 3 AZs
It supports both document and key-value pair
The maximum item size in DynamoDB is 400 KB, which includes both attribute name binary length (UTF-8 length) and attribute value lengths (again binary length).
Explain Eventual Consistency Read VS Strongly Consistency Read
Eventual means that the written data will take up to 1s to be available on the replicas
Strong consistency read: This is recommended to be enabled in case the application needs to read in less than 1s
What is DynamoDB Accelerator (DAX)?
It is the in-memory cache in front of DynamoDB that improves read/write performance up to 10x while not requiring any code change as it is fully compatible with the DynamoDB API
How DynamoDB supports transactions?
It prepares and then commits, so your are charged for more capacity
It supports up to 25 items or 4MB of data per transaction
Interesting to know that DMS (Database Migration Service) supports DynamoDB/Aurora/Kafka/S3/Kinesis/Elasticsearch as target
How DynamoDB is charged?
Provisioned: Per request and storage
OnDemand: Pay-per-request, but requests are more expansive than provisioned. As the application settle down, it is a good idea to migrate to provisioned, and use VPC Endpoints to save money
How DynamoDB backup works?
Same region full backups anytime with zero impact. The retention is up to 35 days;
Point-in-time recovery is not enabled by default, but it permits RPO of 5min.
How DynamoDB global tables work?
Multi-master; Multi-region; Less than 1s replication latency over regions for new items; Based on DynamoDB streams; No need to rewrite the application;
