interactive deployment diagram
- Java w/o any framework
- libraries:
- build tool:
|-- buildSrc
- gradle configuration files for the multi-module project
|-- src/main/groovy/tweets.backend.java-*-conventions.gradle
- base gradle configuration files for all projects/subprojects
- 3 types of files:
- application -
- library - library / shared project configuration
- common - base configuration for both application and library type projects
|-- functions
- each subproject is a Lambda Function
|-- authorizer
- Lambda-based API Gateway authorizer
- handles JWT validation for requests to protected resources
|-- sign-in
- generates JWTs for users
|-- get-tweets
- handles querying for Tweets
|-- create-tweets
- handles creating Tweets
|-- shared
- common library-type projects that are imported and used in Lambda Functions
|-- jwt
- utilities for generating and validating JWTs
|-- rest
- utilities for handling
|-- persistence
- database (DynamoDB) connectivity; handles querying and insertion of data
|-- ...
Based on this excellent talk by Alex Debrie: AWS re:Invent 2019: Data modeling with Amazon DynamoDB (CMY304)
Identified entities:
- User:
| PK (Primary Key) | SK (Primary Key) | CreatedAt | PasswordHash | ProfilePictureURL | Username | |
|---|---|---|---|---|---|---|
| Explanation | Partition Key | Sort Key (PROFILE#) | ISO8601 string | encrypted password | User profile picture | |
| Example | USER#fszuberski | PROFILE#fszuberski | PROFILE#fszuberski | (bcrypt) | https://i.pinimg.com/originals/18/ed/33/18ed330d561a580f77dc39c941455f0e.jpg | fszuberski |
- Tweet:
| PK (Primary Key) | SK (Primary Key) | CreatedAt | CreatedBy | ProfilePictureURL | Text | TweetId | |
|---|---|---|---|---|---|---|---|
| Explanation | Partition Key | Sort Key (TWEET##) | ISO8601 string | User identifier | User profile picture | Unique Tweet UUID | |
| Example | USER#fszuberski | TWEET#2021-05-22T14:43:18Z#2d07d914-e3d8-4f74-a5dc-c7fcc053a710 | 2021-05-31T19:08:28.422847 | USER#fszuberski | https://i.pinimg.com/originals/18/ed/33/18ed330d561a580f77dc39c941455f0e.jpg | Calling it "serverless" without acknowledging the servers is like calling it "passive income" without acknowledging the effort to make it "passive" | 2d07d914-e3d8-4f74-a5dc-c7fcc053a710 |
Identified queries:
-
fetch tweets created by user
-
fetch tweets created by other users
-
(additional) fetch all users
- currently is a full table scan
- can be easily turned into a query by extending the challenge
- extention: handling of user followers and creating tweets only for the followers
- solution: keeping a list
-
Additional notes on current design:
- a copy of each Tweet is created for every User (including the User creating the Tweet)
- optimised for reads
Tech stack: