This paper addresses two unresolved issues about Web caching. The first issue is whether requests from a fixed user community are distributed according to Zipf's (1929) law. second relates number of studies on the characteristics proxy traces, which have shown that hit-ratios and temporal locality traces exhibit certain asymptotic properties uniform across different sets traces. In particular, question these inherent accesses or they simply an artifact An answer will facilitate both cache...

The sharing of caches among Web proxies is an important technique to reduce traffic and alleviate network bottlenecks. Nevertheless it not widely deployed due the overhead existing protocols. In this paper we demonstrate benefits cache sharing, measure protocols, propose a new protocol called "summary cache". protocol, each proxy keeps summary directory participating proxy, checks these summaries for potential hits before sending any queries. Two factors contribute our protocol's low...

Decentralized and unstructured peer-to-peer networks such as Gnutella are attractive for certain applications because they require no centralized directories precise control over network topology or data placement. However, the flooding-based query algorithm used in does not scale; each generates a large amount of traffic systems quickly become overwhelmed by query-induced load. This paper explores, through simulation, various alternatives to Gnutella's algorithm, replication strategy,...

The sharing of caches among Web proxies is an important technique to reduce traffic and alleviate network bottlenecks. Nevertheless it not widely deployed due the overhead existing protocols. In this paper we propose a new protocol called "Summary Cache"; each proxy keeps summary URLs cached documents participating checks these summaries for potential hits before sending any queries. Two factors contribute low overhead: are updated only periodically, representations economical --- as 8 bits...

Early diagnosis of gear transmission has been a significant challenge, because faults occur primarily at microstructure or even material level but their effects can only be observed indirectly system level. The performance fault depends significantly on the features extracted and classifier subsequently applied. Traditionally, fault-related are identified based domain expertise through data preprocessing which system-specific may not easily generalized. On other hand, although recently deep...

Peer-to-peer (P2P) applications such as BitTorrent ignore traffic costs at ISPs and generate a large amount of cross-ISP traffic. As result, often throttle to control the cost. In this paper, we examine new approach enhance locality, biased neighbor selection, in which peer chooses majority, but not all, its neighbors from peers within same ISP. Using simulations, show that selection maintains nearly optimal performance Bit- Torrent variety environments, fundamentally reduces by eliminating...

As the Web continues to explode in size, caching becomes increasingly important. With comes problem of cache consistency. Conventional wisdom holds that strong consistency is too expensive for Web, and weak methods, such as Time-To-Live (TTL), are most appropriate. This study compares three approaches: adaptive TTL, polling-every-time invalidation, through analysis, implementation, trace replay a simulated environment. Our analysis shows methods save network bandwidth mostly at expense...

Prefetching and caching are effective techniques for improving the performance of file systems, but they have not been studied in an integrated fashion. This paper proposes four properties that optimal strategies prefetching must satisfy, then presents studies two such strategies, called aggressive conservative. We prove conservative approach is within a factor strategy significantly less than twice case. evaluated these approaches by trace-driven simulation with collection access traces....

This paper presents a new algorithm to answer top-k queries (e.g. find the k objects with highest aggregate values) in distributed network. Existing algorithms such as Threshold Algorithm [10] consume an excessive amount of bandwidth when number nodes, m, is high. We propose called Three-Phase Uniform (TPUT). TPUT reduces network consumption by pruning away ineligible objects, and terminates three round-trips regardless data input.The two sets results about TPUT. First, trace-driven...

Article Free Access Share on Web prefetching between low-bandwidth clients and proxies: potential performance Authors: Li Fan Department of Computer Science, University Wisconsin-Madison Wisconsin-MadisonView Profile , Pei Cao Wei Lin Quinn Jacobson Electrical Engineering, Authors Info & Claims SIGMETRICS '99: Proceedings the 1999 ACM international conference Measurement modeling computer systemsMay 1999Pages 178–187https://doi.org/10.1145/301453.301557Published:01 May 1999Publication...

As the performance gap between disks and micropocessors continues to increase, effective utilization of file cache becomes increasingly immportant. Application-controlled caching prefetching can apply application-specific knowledge improve management. However, supporting application-controlled is nontrivial because need be integrated carefully, kernel needs allocate blocks among processes appropriately. This article presents design, implementation, a system that integrates caching,...

Reducing energy consumption is an important issue for data centers. Among the various components of a center, storage one biggest consumers energy. Previous studies have shown that average idle period server disk in center very small compared to time taken spin down and up. This significantly limits effectiveness power management schemes. paper proposes several power-aware cache algorithms provide more opportunities underlying schemes save More specifically, we present off-line greedy...

Dynamic documents constitute an increasing percentage of contents on the Web, and caching dynamic becomes increasingly important issue that affects scalability Web. In this paper, we propose Active Cache scheme to support at Web proxies. The allows servers supply cache applets be attached with documents, requires proxies invoke upon hits furnish necessary processing without contacting server. We describe protocol, interface security mechanisms scheme, illustrate its use via several examples....

As the Web continues to explode in size, caching becomes increasingly important. With comes problem of cache consistency. Conventional wisdom holds that strong consistency is too expensive for Web, and weak methods such as Time-To-Live (TTL) are most appropriate. The article compares three approaches: adaptive TTL, polling-every-time, invalidation, using prototype implementation trace replay a simulated environment. Our results show invalidation generates less or comparable amount network...

Decentralized and unstructured peer-to-peer networks such as Gnutella are attractive for certain applications because they require no centralized directories precise control over network topology or data placement. However, the flooding-based query algorithm used in does not scale; each individual generates a large amount of traffic systems quickly become overwhelmed by query-induced load. This paper explores various alternatives to Gnutella's replication strategy. We propose based on...

Prefetching and caching are effective techniques for improving the performance of file systems, but they have not been studied in an integrated fashion. This paper proposes four properties that optimal strategies prefetching must satisfy, then presents studies two such strategies, called aggressive conservative. We prove conservative approach is within a factor strategy significantly less than twice case. evaluated these approaches by trace-driven simulation with collection access traces....

Traditional file system implementations do not allow applications to control caching replacement decisions. We have implemented two-level replacement, a scheme that allows their own cache while letting the kernel allocation of space among processes. designed an interface let exert on via set directives kernel. This is effective and requires low overhead.We demonstrate for perform well under traditional policies, combination good application-chosen strategies, our policy LRU-SP, can reduce...

As disk performance continues to lag behind that of memory systems and processors, virtual management becomes increasingly important for overall system performance. In this paper we study the page reference behavior a collection memory-intensive applications, propose new replacement algorithm, SEQ. SEQ detects long sequences faults applies most-recently-used those sequences. Simulations show large class performs close optimal significantly better than Least-Recently-Used (LRU). addition,...

No abstract available.

The sharing of caches among Web proxies is an important technique to reduce traffic and alleviate network bottlenecks. Nevertheless it not widely deployed due the overhead existing protocols. In this paper we propose a new protocol called "Summary Cache"; each proxy keeps summary URLs cached documents participating checks these summaries for potential hits before sending any queries. Two factors contribute low overhead: are updated only periodically, representations economical --- as 8 bits...