T. P. Kelly, S. Jarmin and J. K. MacKie-Mason, "Variable QoS from Shared Web Caches: UserCentered Design and Value Sensitive Replacement," Proc. ISQE, Dec 1999.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....of cache space is allocated to the publisher. In order to understand the form of publisher utility functions, we need to consider the user interests for web objects. It has been shown that users are most interested in a small portion of the content in the web server, for example see [5] 6] and [7]. These studies have determined that user interests for web objects can be realistically modeled by Zipf distribution [11] Zipf distribution suggests that the popular objects are requested exponentially more than the less popular objects. Intuitively, if the object sizes are the same the ....

T. P. Kelly, S. Jarmin and J. K. MacKie-Mason, "Variable QoS from Shared Web Caches: UserCentered Design and Value Sensitive Replacement," Proc. ISQE, Dec 1999.

....to natural but not fully rational bidding strategies (Wellman et al. to appear) Another method is to simulate allocation outcomes under reasonable assumptions on agent behavior. We have used this method to assess allocation efficiency (Anderson, Birgean, and MacKie Mason 1999; Wurman 1999)(Kelly et al. 1999a, 1999b; Callaway or other packet scheduling work) We have used both of these methods to assess the performance of a non price economic allocation mechanism (the Generalized Vickrey Auction (MacKie Mason and Varian 1994) For example, in our application of the GVA to scheduling (Wellman et al. ....

....cache serves u. Experiments have shown that this simple generalization of LFU delivers substantially higher aggregate value to servers than ordinary LFU for some distributions of V u . For example, Figure 2 presents simulation results for a two week interval of NLANR data for a large cache site (Kelly et al. 1999). Our performance metric, value hit rate, measures the fraction of request value delivered to users, where value is given by the weights V u . As the graph shows, weighted LFU outperforms LFU or LRU (the least recently used replacement policy) for any given cache size. Weighted LFU performs even ....

Kelly, Terence P., Sugih Jamin, and Jeffrey K. MacKie-Mason. 1999. Variable QoS from Shared Web Caches: User-Centered Design and Value-Sensitive Replacement.

....Because it allows us to compute A (s) at the additional cost of sorting the output, in effect this algorithm enables us to simulate all cache sizes of possible interest simultaneously. An efficient method is necessary in order to process real traces, in which M and N can both exceed 10 million [27]. To make the issue concrete, whereas a na ve O(MN) priority depth algorithm required over five days to process 11.6 million requests for 5.25 million documents, our O(M log N) algorithm completed the job in roughly 3 min on the same computer. In order for our method to work, we require that the ....

....Perfect LFU is optimal for infinite sequences of independent references from a fixed distribution, if document sizes are uniform. Limited empirical evidence, however, suggests that optionalplacement variants of LFU perform worse than their mandatory placement counterparts on real Web workloads [27]; the subject has not been investigated thoroughly. GD Size [14] and mandatory placement variants of LFU such as GDSF [4] swLFU [26,27] and LUV [8] do not satisfy the inclusion property, and therefore the one pass simulation methods described in Section 3.1 cannot be applied to them. 3.2. ....

[Article contains additional citation context not shown here]

T. Kelly, S. Jamin, J.K. MacKie-Mason, Variable QoS from shared Web caches: user-centered design and value-sensitive replacement, Proceedings of the MIT Workshop on Internet Service Quality Economics (ISQE 99), Cambridge, MA, December 1999, http:// www.marengoresearch.com/isqe/agenda_m.htm.

....Because it allows us to compute A (s) at the additional cost of sorting the output, in effect this algorithm enables us to simulate all cache sizes of possible interest simultaneously. An efficient method is necessary in order to process real traces, in which M and N can both exceed 10 million [27]. To make the issue concrete, whereas a nave O(MN) priority depth algorithm required over five days to process 11.6 million requests for 5.25 million documents, our O(M log N) algorithm completed the job in roughly three minutes on the same computer. In order for our method to work, we require ....

....Perfect LFU is optimal for infinite sequences of independent references from a fixed distribution, if document sizes are uniform. Limited empirical evidence, however, suggests that optionalplacement variants of LFU perform worse than their mandatory placement counterparts on real Web workloads [27]; the subject has not been investigated thoroughly. GD Size [14] and mandatory placement variants of LFU such as GDSF [4] swLFU [26, 27] and LUV [8] do not satisfy the inclusion property, and therefore the one pass simulation methods described in Section 3.1 cannot be applied to them. 7 Table ....

[Article contains additional citation context not shown here]

Terence Kelly, Sugih Jamin, and Jeffrey K. MacKie-Mason. Variable QoS from shared Web caches: User-centered design and value-sensitive replacement. In Proceedings of the MIT Workshop on Internet Service Quality Economics (ISQE 99), Cambridge, MA, December 1999. http://www.marengoresearch.com/isqe/agenda_m.htm.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC