Abstract- Caching systems try to manage and balance memory usage and speed by relying on eviction and promotion policies; however, most web developers find it obscure how cached data moves behind the scenes. This paper presents the design and evaluation of a twotier cache system consisting of a HOT in-memory layer and a distributed COLD layer. In this system, old data is eliminated by TTL-based lazy-eviction, and data is immediately promoted from the cold-tier to the hot-tier upon access. To understand this in its entirety, an observability dashboard was developed to showcase hit patterns, TTL states, and promotion events in real time. Through controlled workload experiments, we observed how lazy eviction and promotion rules affect cache hit rates and data churn. The results demonstrate visualized cache behavior in real time, providing a detailed understanding of how eviction and promotion affect cache performance.
