Project Structure

• The project consists of the following microservices.
  • Gateway
  • Server
  • User
  • Location-Writer
  • Location-Reader
• Instructions on how to run these projects is given in their respective README.md
• Eureka is used for microservice discovery along with Spring Cloud for loadbalancing.
• The service Location-Writer is used to recieve location data. Since there will be a lot of request coming to this end point I've created multiple instances of this service which are loadbalanced.
• The loadbalancing startegy is the default Round Robbin, I've also created a custom strategy based on URL hash for server stickiness. The reason for this was because we have multiple instances of Location microservice, the data of a particular user would be distributed, making pagination and data retrieval complex.
• However my custom URL hash load balancer depends on number of instances for server stickiness, so if instances increases the data would be distributed again.
• Keeping this in mind I created another microservice Location-Reader which would sync with the Location-Writer.
• The Location-Reader's database would have partitions as well as indexes for fast retrival.
• These two databases sync with each other by an Async method that send location data through kafka from Location-Writer to Location-Reader. (Previously I used Debezium but that seemed over kill, as well as a few problems I faced with Debezium connector).
• In Location-Reader Database Location is partitioned based on user_id column for fast retrieval since a user can have thousands of entries.
• created_on column has been indexed as well for doing time frame query.
• The id column is an auto-increment column so the latest location of the user is the last value inserted in the partition.