TargetingEngineGG

TargetingEngineGG

This repository implements a targeting engine that routes campaigns to the right requests, with a focus on extensible, maintainable, and reusable code.

Database Setup

1. Start MySQL

   brew services start mysql
   
   

2. Create the database and tables

   mysql -u root -p < schema.sql
   
   

Scalable Targeting Logic

Targeting logic and dimension handling are decoupled, so you can add new dimensions easily without changing application code.

• Dimension is stored as a VARCHAR, not an ENUM.

  • ENUM requires a schema change every time you add a new dimension.

  • VARCHAR allows arbitrary dimension names without DB migration.

  INSERT INTO targeting_rules (campaign_id, dimension, type, value)
  VALUES ('spotify', 'DeviceType', 'INCLUDE', 'Tablet');
  
  

Targeting Dimensions

Adding a New dimension

1. Update the RequiredDimensions or OptionalDimensions list in delivery/config.go.

2. Add the corresponding targeting rules in the DB:

   INSERT INTO targeting_rules (campaign_id, dimension, type, value)
   VALUES ('spotify', 'SubscriptionTier', 'INCLUDE', 'Premium');
   
   

Handling the read heavy workload

1. In a service like this, if there are ~1000s of campaigns, thats KBs of data. And with 100000 of requests, thats still MBs of data. Adding an In-memory cache using the service's memory would be the fastest way to cache, as it will avoid the network hop required to query redis or anything.

2. But Since I am trying to make a production ready service, such a service can be deployed on multiple pods, which can be sharing cache. Its better to implement redis, which is what I will do :)