The client requests for the merchant data from the backend - which is first redirected to the optimal instance by the load balancer.
The request is queued for the server to respond to.
The central server contacts the cache server (L2 caching) - and if found instantly returns the merchant data along with ± 10 pincodes being cached on the client itself (L1 caching).
The L1 cache covers almost 90% of real use cases. As the user will always want to see nearby pincode data. Which will be cached on the first request itself.
If data isn’t found in the cache server, the central server contacts the data base managers for data
The request is directed to the database managers, which fetch the data from the database. We have implemented horizontal db sharing based on geolocation.
The response is shared with the central server which is redirected to the client. The db queues the caching of ±50 nearest pincodes to the cache server (L2 cache)
L2 cache grows in size as more and more users request data.
Uploading the data
The client upload the Excel data of the merchants and pincodes in this format
The request is queued and the data is streamed onto the database.
