Recommendation Cache Fallback for Server-Fault Content Delivery
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The stability of recommendation information flow applications is heavily dependent on the stability of the recommendation system, leading to random content provision when the system fails, which negatively affects user experience.
Innovation Solution
A content recommendation method and apparatus that includes caching recommendation content items when the server is functioning normally and using these cached items when the server fails, along with a mechanism to determine server status and manage cache capacity based on user activity and server performance metrics.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the recommendation system is highly dependent on the recommendation server, then the recommendation accuracy can be maintained, but the system stability deteriorates when the server fails
Solution Approach 1:
The system performs preliminary actions by pre-calculating and caching recommendation content sequences before the server fails. The information flow server stores multiple recommendation content sequences in advance, so when the recommendation server becomes unavailable, it can immediately serve cached content without interruption, maintaining system stability and content delivery continuity.
Solution Approach 2:
The information flow server acts as an intermediary between the terminal device and the recommendation server. It caches recommendation content sequences locally and can provide content to terminal devices independently when the recommendation server fails, serving as a mediator that ensures continuous service availability.
2Reliability
If multiple recommendation content sequences are cached for each user, then the content delivery continuity is improved during server failures, but the cache space consumption increases
Solution Approach 1:
The system applies partial action by caching a limited number of recommendation content sequences (e.g., 3 sequences) for each user instead of caching all possible content. This partial caching provides sufficient continuity during server failures while controlling cache space consumption within acceptable limits.
3Speed
If the system caches recommendation content sequences in advance, then the response speed during server failures is improved, but the memory resource consumption increases
Solution Approach 1:
The system caches a partial set of recommendation content sequences (typically 3 sequences per user) in advance rather than all possible content. This approach achieves fast response during server failures by having pre-computed content ready, while controlling memory resource consumption through selective caching of only the most relevant sequences.
Data Source
AI summary
This disclosure discloses a content recommendation method and apparatus. The content recommendation method includes: receiving, from a terminal device, content request information for requesting to acquire a recommendation content item corresponding to a target user; determining whether a recommendation server is faulty; acquiring a first recommendation content item according to a recommendation cache corresponding to the target user in response to the recommendation server being faulty, wherein a plurality of recommendation content items are stored in the recommendation cache corresponding to the target user, and the plurality of recommendation content items are written when the recommendation server is not faulty; and sending the first recommendation content item to the terminal device.