Abstract not found

Popular documents are frequently mirrored on multiple sites in an effort to share the load and reduce clients' retrieval latencies. However, choosing the best mirror site is a non-trivial task and a bad choice may give poor performance. We propose a scheme in which clients access multiple mirror sites in parallel to speedup document downloads while eliminating the problem of server selection. In our scheme, clients connect to mirror sites using unicast TCP connections and dynamically request different pieces of a document from different sites. The amount of data retrieved from a particular site varies depending on the network path/server conditions. Dynamic parallel-access can be easily implemented in the current Internet and does not require any modifications at the mirror sites. Using dynamic dynamic parallel-access, all clients experience dramatic speedups in downloading documents, and the load is shared among servers without the need for a server selection mechanism. Even in a situ...

Abstract. We revisit the problem of selecting the suitable server, among a list of replica servers, so as to optimize the response time resulting from the execution of the clients ’ requests. Our proposition, presented in this paper, consists in the definition of a function estimating the response time and a method for applying it to different application workloads. The function combines the application demands for various resources (such as the CPU, the disk I/O and the network bandwidth) with the resource capabilities and availabilities on the replica servers. The main benefits of our approach include: the simplicity and the transparency, from the perspective of the clients, who don’t have to specify themselves the selection criteria, the estimation accuracy, by considering the application real needs and the current degree of resource usage, determined by concurrent applications and the flexibility, with respect to the precision with which the resource-concerned parameters are specified. The experiments we conducted show two positive results. Firstly, our estimator provides a good approximation of the real response time obtained by measurements. Secondly, the ordering of the servers according to our estimation function values, matches with high accuracy the ordering determined by the real response times. 1

Abstract not found

The rapid growth of the Internet in users and content has fueled extensive efforts to improve the user's overall Internet experience. A growing number of providers deliver content from multiple servers or proxies to reduce response time by moving content closer to end users. An increasingly popular mechanism to direct clients to the closest point of service is DNS-based redirection, due to its transparency and generality. This paper draws attention to two of the main issues in using DNS: 1) the negative effects of reducing or eliminating the cache lifetimes of DNS information, and 2) the implicit assumption that client nameservers are indicative of actual client location and performance. We quantify the impact of reducing DNS TTL values on web access latency and show that it can increase name resolution latency by two orders of magnitude. Using HTTP and DNS server logs, as well as a large number of dial-up ISP clients, we measure client-nameserver proximity and show that a significant fraction are distant, more than # hops apart. Finally, we suggest protocol modifications to improve the accuracy of DNS-based redirection schemes. I.