2

I have to create a Service, Which has to persist a flag against a userId using the api void register(String userId) and also the service has to vend a api boolean isRegistered(String userId) which says whether a UserID has registered for the service or not.. I’m planning to have a No-SQL database as a Backend as reads and writes are O(1) I prefer not to change the db store as it is vended by a different service.

Expected Request Per Second for create api is : 10 (10 writes per sec to db)

Expected Request Per Second for isRegistered api is : 10000 (10000 reads per second to db). which means for more than 95% of the request this API will return false and userId for isRegistered are mostly distinct values, So caching would need a more ttl to reap its benefit.

Now I’m trying to optimize number of database reads made by isRegistered() call as most of the values returned by this api are going to be false(95% of the time).

What is the best way to design this System/DB. Is there an existing design pattern which I can use to Solve this Problem? like a better way to cache.

Thanks in Advance

VINOTHKUMAR MOHAN
  • 21
  • 3
  • "What is the best way to.. " this question is primarily opinion based.. Take a look at cache aside pattern: https://blog.cdemi.io/design-patterns-cache-aside-pattern/ – Oscar Mar 05 '18 at 15:47
  • Thanks , I'm looking for a Design pattern which is suitable for this kind of problem – VINOTHKUMAR MOHAN Mar 08 '18 at 05:50

0 Answers0