SILC: Lookahead Caching for Short-form Video Delivery Systems
2026-05-06 • Networking and Internet Architecture
Networking and Internet Architecture
AI summaryⓘ
The authors study how popular short video apps like TikTok deliver videos efficiently. They notice that because these apps push videos to users in a sequence and some videos are very popular, it's possible to predict what videos will be requested soon. Using this idea, they create SILC, a caching system that stores videos more smartly to reduce delays and data transfer costs for content networks. They tested SILC with real user data and found it works better than existing methods, cutting down network costs significantly. Their work helps make streaming short videos smoother and cheaper.
short video platformspush-based recommendationcontent delivery network (CDN)cache miss ratesmidgress bandwidthPareto distributioncache eviction policiesSILCuser trace data
Authors
Maleeha Masood, Shreya Kannan, Om Chabra, Deepak Vasisht, Indranil Gupta
Abstract
Short video platforms like TikTok, Instagram Reels, and YouTube Shorts have gained immense popularity in the last few years and are responsible for a large and growing fraction of Internet traffic. We identify two unique opportunities for improving short video delivery using their existing interactions with content delivery networks (CDNs). First, short videos use a push-based recommendation system, where the user is presented a sequence of videos recommended by the algorithm rather than user explicitly picking content to watch (e.g., in YouTube). Such push-based short video systems offer a unique opportunity for system design by providing visibility into upcoming requests. Second, the popularity of these videos follows a highly skewed Pareto distribution, leading to geographical and temporal overlap amongst videos being served. We leverage these opportunities to build SILC - a lookahead-aware caching system, aimed at (i) reducing CDN cache miss rates, as well as (ii) reducing midgress bandwidth between the CDN and the origin server. Our evaluation of SILC uses traces that we collect from real users, through (i) an in-person user study, and (ii) a data donation program involving 100 TikTok users across the world. Using a combination of these traces, we simulate traffic from 10,000 simultaneous users. Our evaluation shows that, compared to 10 state-of-the-art heuristic and learning-based cache eviction policies, SILC reduces a CDN's midgress costs by 11.1% to 111%.