One service layer needs to call the other two service layers to obtain data and assemble it into the final data. The data is all lists. How to design the cache?

The scenario is as follows: the A-service layer needs to call the B-service layer to obtain a data B-list, and then use the obtained B-list to call the C-sevice layer to obtain the data C-list, and finally assemble the B-listGet the A-list with the data of the C-list and return it.
Question:
1. Should the A-list be added to the redis cache, or should the B-list and C-list be added to the cache?
2. With the increase of cache keys, when an entity data is modified, it is necessary to consider the invalidation of the cached keys involved, which is a headache. It is necessary to add a new key invalidation code in the updated place, andI have to invalidate multiple keys. Do you have this kind of trouble?
Just started work, looking for answers!

---

Other Answer 1:

Why not make cache through in B-service and C-sevice?

Why at A-service?What is the point of doing?Where are the bottlenecks?