CAP Theorem in Distributed Systems

You are asked to design a distributed key-value store that must continue to serve read and write requests when a network partition occurs between two data-centers. The store is the source-of-truth for (1) a user’s current account balance used by a trading desk and (2) the “last-price” cache displayed on a market-data dashboard. During a partition the link between DC-East and DC-West is lost; each side can talk to its local clients and to its local replica, but not to the other replica. Explain, for each use-case, whether your system should reject requests, return possibly stale data, or take another approach. Justify your choice in terms of the CAP theorem, describe the exact user-visible behavior when the partition happens, and sketch the minimal mechanism (quorum, versioning, conflict-resolution, etc.) required to implement your decision. You do not need to code; a clear architectural explanation with trade-offs is expected.