Asymptotically Tight Approximation for Online File Caching With Delayed Hits and Bypassing
In latency-sensitive file caching systems such as Content Delivery Networks (CDNs) and Mobile Edge Computing (MEC), the latency of fetching a missing file to the local cache can be significant. Recent studies have revealed that successive requests for the same missing file before the fetching process completes could still suffer latency (so-called delayed hits). Motivated by the practical scenarios, we study the online general file caching problem with delayed hits and bypassing, <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">i.e.</i>, a request may be bypassed and processed directly at the remote data center. The objective is to minimize the total request latency. We present a general reduction that turns a traditional file caching algorithm into one that can handle delayed hits. Based on this reduction, we propose an efficient online file caching algorithm, called <monospace xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">CaLa</monospace>, with an asymptotically tight competitive ratio as <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$O(Z \log K)$ </tex-math></inline-formula>, where <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Z</i> is the maximum fetching latency of any file and <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">K</i> is the cache size. Extensive simulations on the production data trace from Google and the Yahoo benchmark illustrate that <monospace xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">CaLa</monospace> can reduce the latency by up to 8.48% compared with the state-of-the-art schemes dealing with delayed hits without bypassing, and this improvement increases to 26.00% if bypassing is allowed. Furthermore, by upgrading the method for estimating files’ weights in <monospace xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">CaLa</monospace>, we propose <monospace xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">CaLa+</monospace>, which further reduces the total latency by more than 5%.
Authors
- Zhenhua Han (ORCID: https://orcid.org/0000-0002-2880-7100)
- Haohua Du (ORCID: https://orcid.org/0000-0002-8492-3990)
- Shaofeng H.-C. Jiang (ORCID: https://orcid.org/0000-0001-7972-827X)
- Haisheng Tan (ORCID: https://orcid.org/0000-0002-3133-1430)
- Chi Zhang (ORCID: https://orcid.org/0000-0003-1160-5497)
- Xiang‐Yang Li (ORCID: https://orcid.org/0000-0002-6070-6625)
- Guopeng Li (ORCID: https://orcid.org/0000-0002-0713-8964)
- Yi Wang (ORCID: https://orcid.org/0009-0004-0665-5945)
Institutions
- University of Science and Technology of China (CN)
- Hefei University of Technology (CN)
- Peking University (CN)
- Microsoft Research Asia (China) (CN)
- Beihang University (CN)
Publication Details
- Journal
- IEEE Transactions on Networking
- Published
- 2025-03-24
- DOI
- https://doi.org/10.1109/ton.2025.3549289
- Citations
- 1
- Primary Topic
- Caching and Content Delivery
- Type
- article
- Field-Weighted Citation Impact
- 1.49
Funders
- National Natural Science Foundation of China