Choosing datastores

Creating, editing, deleting and querying data related to merchant partners of ZestMoney and what EMI Financing options to allow for them was a major problem we solved at Zest. While doing a deeper analysis of the problem, we realised that business data entry and querying systems usually have different functional and non-functional requirements. The data creation and related updates are done by business users, usually during business hours. It is highly structured and interconnected data with inter-relationship among entities. While the data modelling of such systems is a challenge, it does not have to be highly available and is not expected to handle high loads. The querying system on the other hand, is required to have very low response times, scale to high loads and be highly available. What is intuitively CRUD, is practically two different sets of requirements, responsibilities and performance. This talk is about how we solved this problem using the CQRS - Command query responsibility segregation - pattern and the interesting data modelling that followed