Imposing Cache: Busy-Aware Cooperative Data Caching in Edge-Cloud Environments
Gyujeong Lim, Jihun Kang, Heonchang Yu · 2023
Data caching and placement are essential for reducing latency and increasing performance. With the correct allocation, users can access frequently used data faster while reducing the load on network bandwidth and resources. However, with the ever changing virtual systems, caching methods must also adapt to fit in them. For example, cloud services no longer operate by only having a single powerful node such as a data center. Instead, services also deploy edge nodes, which can hold data closer to the users, thus reducing latency and improving quality. As a result, there are various studies and developments to make caching more efficient than ever. In this paper, we focus on what we call imposing cache, a method that allows nodes to cache on other nodes as if it were their own, as long as they are part of the same group we call EdgeGroup. Additionally, our method is busy-aware enabling cloud and edge nodes to check on each other’s availability and adjust their decision making process so that unavailable nodes are not taken into consideration. Our method succeeded in reducing fetches from Cloud by around 33% compared to normal caching methods. Also, it reduced request completion time by around 3 minutes due to the fact that it could correctly identify busy nodes and suspend them accordingly.