1.Web-Conscious Storage Management for Web Proxies
2.user-level disk management
3.Using realistic trace-driven simulations we show that the proxy’s
4.secondary storage I/O throughput can be improved by a factor of 18, enabling a single-disk proxy to serve around 500 (URL-get) operations per second.
5.Our experimental evaluation suggests that Foxy outperforms traditional proxies such as SQUID by more than a factor of 7 under heavy load.
6.non-caching mode
7.proof-of-concept proxy server
8.cache miss
9.data-locality
10.Most of the meta-data overhead that cripples web proxy performance can be traced to the storage of each URL in a separate file.
11.To eliminate this performance bottleneck we propose a novel URL-grouping method (called BUDDY), in which we store all the URLs into a small number of files.
12.Towards this end, we develop a file-space management algorithm (called STREAM) that (much-like log-structured file systems) streams write operations to the disk: The web proxy stores all URLs in a single file organized in slots of 512 bytes long. Each URL occupies an integer number of (usually) contiguous slots.
13.LAZY-READS technique
14.NURL
15.consider the case of an HTML page that contains several embedded images.
16.A first step towards improving the performance of read operations, LAZYREADS, reduce this overhead by clustering several read operations and by sending them to the disk together. This grouping of read requests not only reduces the disk head ping-pong effect but also presents the system with better opportunities for disk head scheduling.
发表于 2006-10-10 14:28:02
提升卡
變色卡
千斤頂