KV-Cache: A Scalable High-Performance Web-Object Cache for Manycore

Latency and cost of Internet-based services are driving the proliferation of web-object caching. Memcached, the most broadly deployed web-object caching solution, is a key infrastructure component for many companies that offer services via the Web, such as Amazon, Facebook, Linked In, Twitter, Wikipedia, and YouTube. Its aim is to reduce service latency and improve processing capability on back-end data servers by caching immutable data closer to the client machines. Caching of key-value pairs is performed solely in memory. In this paper, we present a novel design for a high-performance web-object caching solution, KV-Cache, that is Memcache-protocol compliant. Our solution, based on TU Dresden\´s Fiasco. OC micro kernel operating system, offers scalability and performance that significantly exceeds that of its Linux-based counterpart. KV-Cache\´s highly optimized architecture benefits from truly "absolute" zero copy by eliminating any software memory copying at the kernel level or in the network stack, and only performing direct memory access (DMA) for each transmit and receive path. We report experimental results for the current prototype running on an Intel E5-based 32-core server platform. Our results show that KV-Cache offers significant performance advantages over optimized Memcached on Linux for commodity x86 server hardware.