Article: RIPQ: Advanced Photo Caching on Flash for Facebook
Summary: The main aim of the paper is to address the issue of designing a flash cache which is able to support advanced cashing algorithms and can be used in Facebook photo workload. These algorithms help to utilize the maximum capacity of a flash available cache. The researchers have designed and implemented Restricted Insertion Priority Queue (RIPQ) framework for this purpose.
Strengths: Researchers have implemented two advanced caching algorithms to evaluate the flexibility of RIPQ. Both these algorithms demonstrated considerable improvement in cashing performance.
Weaknesses: Majority caching algorithms can be implemented on RIPQ but there are some caching algorithms that cannot be implemented with a single RIPQ e.g. MQ and ARC. The framework design of current study is not capable of implementing these algorithms as it incorporates single RIPQ.
Unsolved Problem/future directions: The current study does not support delete/overwrite. In the future studies these characteristics can also be added. In future a framework can be designed to incorporate multiple RIPQ so that it can be used with advance caching algorithms like MQ and ARC.
Questions: How efficient would be the performance of caching algorithms if a RAM is used instead of Flash memory?