Research System Design

[underwatercolors]

Assumptions:

1. Service is used by a big size retail store like costco.
2. Service is used by 20 cashiers/self-checkout-shoppers at a time.
3. Database is read and write heavy. Read prices, stock number, user info, etc. Write sales transaction, price updates, stock number updates, user points, etc.
4. A cache can be used for prices as it doesn't update often.
5. There can be anonymous session and known session.

frontend

• Angular, React , Vue (Web Client)
• IOS, Android (Mobile Client)
• Wpf (Windows Client)

backend services

• Rest API
• GraphQL
• GRPC

Database

• NoSQL
• SQL

Database Model design? Caching Services?? Serverless Function?? Usage of Docker??

https://app.diagrams.net/#Linventory-pos

[underwatercolors]

NoSQL vs SQL I think SQL is more appropriate because

• our data are structured and are in small amount
• our database model schema is mostly rigid and do not change much
• we prioritize more on consistency than availability
• our need for scalability/distributed database is low

Database Model Design

[underwatercolors]

Caching Services Redis Low priority. Cache can be placed between services and database. Good for retrieving product prices, eg "MilkPrice":2.00", and product quantity, eg "MilkQuantity:150". Some decision to make when implementing cache: Read-Through, Write-Through, Write-Back, Write-Around, and Refresh-Ahead.

[underwatercolors]

Serverless Function AWS Lambda pros:

• saves money: we don't need to host server during slow business and after business hours
• service can scale up and down quicker than a server

cons:

• difficult to test and debug

[underwatercolors]

AWS Things that can be hosted on aws

• database: Amazon RDS
• docker image: Amazon ECR
• running docker images: Amazon ECS or Amazon Kubernetes
• serverless: AWS Lambda
• storage service: Amazon S3
• notification system: Amazon SNS
• email system: Amazon SES