The Influence of Delayed-hit on Caching Servers Utilizing Conventional Cache Replacement Methods
Feri Fahrianto, Noriaki Kamiyama · 2023
A caching system implemented through online methods like Content Delivery Network (CDN), Proxy, Gateway, or Information-centric Networking (ICN) router is utilized to improve network connectivity. This system utilizes the caching mechanism to store in short-term of data packets obtained from the actual content provider. When the client seeks to access the desired data content, it sends a request to a caching server. Delayed-hit caching, defined as the decline in hit ratio, can be influenced by both the time delay discrepancy between the client’s request packet and the original server’s data packet, as well as the cache’s capacity. This situation arises when the original server’s data packet’s average response time is slower than the average request time of the client’s request packet. Consequently, this paper examines the repercussions of delayed-hit caching in the caching server. It compares two well-known cache replacement algorithms, namely, Least Recently Used (LRU) and First in First out (FIFO). Our findings are proven through simulations conducted with various values of request skewness $(\alpha)$, which indicate the popularity of the requested content following a Zipf distribution against the arrival request time and the response time of the original server.