Huican Zhu

Demand-driven service differentiation in cluster-based network servers

Service differentiation that provides prioritized service qualities to multiple classes of client requests can effectively utilize available server resources. This paper studies how demand-driven service differentiation in terms of end-user performance can be supported in cluster-based network servers. Our objective is to deliver better services to high priority request classes without over-sacrificing low priority classes. To achieve this objective, we propose a dynamic scheduling scheme, called DDSD that adapts to fluctuating request resource demands by periodically repartitioning servers. This scheme also employs priority-based admission control to drop excessive user requests and achieve soft performance guarantees. For each scheduling period, our scheme monitors the system status and uses a queuing model to approximate server behaviors and guide resource allocation. Our experiments show that the proposed technique achieves demand-driven service differentiation while maximizing resource utilization and that it can substantially outperform static server partitioning.

Class-based cache management for dynamic Web content

Caching dynamic pages at a server site is beneficial in reducing server resource demands and it also helps dynamic page caching at proxy sites. Previous work has used fine-grain dependence graphs among individual dynamic pages and underlying data sets to enforce result consistency. This paper proposes a complementary solution for applications that require coarse-grain cache management. The key idea is to partition dynamic pages into classes based on URL patterns so that an application can specify page identification and data dependence, and invoke invalidation for a class of dynamic pages. To make this scheme time-efficient with small space requirement, lazy invalidation is used to minimize slow disk accesses when IDs of dynamic pages are stored in memory with a digest format. Selective precomputing is further proposed to refresh stale pages and smoothen load peaks. A data structure is developed for efficient URL class searching during lazy or eager invalidation. This paper also presents design and implementation of a caching system called Cachuma which integrates the above techniques, runs in tandem with standard Web servers, and allows Web sites to add dynamic page caching capability with minimal changes. The experimental results show that the proposed techniques are effective in supporting coarse-grain cache management and reducing server response times for tested applications.

Scheduling optimization for resource-intensive Web requests on server clusters

Clustering support with a single-system image for large-scale Web servers is important to improve the system scalability in processing a large number of concurrent requests from Internet, especially when those requests involve resource-intensive dynamic content generation. This paper proposes scheduling optimization for a Web server cluster with a master/slave architecture which separates tatic and dynamic content processing. Our experimental results show that the proposed optimization using reservation-based scheduling can produce up to a 68% performance improvement.

Adaptive Algorithms for Cache-Efficient Trie Search

In this paper, we present cache-efficient algorithms for trie search. There are three key features of these algorithms. First, they use different data structures (partitioned-array, B-tree, hashtable, vectors) to represent different nodes in a trie. The choice of the data structure depends on cache characteristics as well as the fanout of the node. Second, they adapt to changes in the fanout at a node by dynamically switching the data structure used to represent the node. Third, the size and the layout of individual data structures is determined based on the size of the symbols in the alphabet as well as characteristics of the cache(s). We evaluate the performance of these algorithms on real and simulated memory hierarchies. Our evaluation indicates that these algorithms outperform alternatives that are otherwise efficient but do not take cache characteristics into consideration. A comparison of the number of instructions executed indicates that these algorithms derive their performance advantage primarily by making better use of the memory hierarchy.

Adaptive load sharing for clustered digital library servers

Abstract.This paper investigates load balancing strategies for clustered Alexandria Digital Library (ADL) servers. The ADL system, which provides online information searching and browsing of spatially-referenced materials through the World Wide Web, involves intensive database I/O and heterogeneous CPU activities. Clustering servers can improve the scalability of the ADL system in response to a large number of simultaneous access requests. One difficulty addressed is that clustered workstation nodes may be non-uniform in terms of CPU and I/O speeds. An optimization scheme is proposed in this paper to dynamically monitor the resource availability, use a low-cost communication strategy for updating load information among nodes, and schedule requests based on both I/O and computation load indices. Since the accurate cost estimation for processing database-searching requests is difficult, a sampling and prediction scheme is used to identify the relative efficiency of nodes for satisfying I/O and CPU demands of these requests. A set of experiments using the ADL traces have been conducted to verify the effectiveness of the proposed strategies.

Hierarchical resource management for Web server clusters with dynamic content

Dynamic content generation has become popular at many Web sites because it enables new services such as electronic commerce, personalized information searching/presentation, and scientific computing. Because dynamic content generation normally places intensive I/O and CPU demands on a server [4, 7], it becomes critical to cluster multiple servers to remove the server bottleneck. One popular method for clustering is to use DNS rotation. The main weakness is that load imbalance may be caused by client-site IP address caching and heterogeneous workload. Also, a client site cannot be aware if a server node fails. A strategy to improve fault tolerance is to use load balancing switching products from Cisco and other companies. The previous DNS and switch based approaches normally use a flat architecture which treats all server nodes equally. Such a solution does not provide a convenient way to dynamically recruit idle resources in handling peak load. Also, switches use simple load balancing schemes which may not be sufficient for resource-intensive dynamic content. Recently, a hierarchical architecture has been proposed for general network services with better flexibility in resource management and fault tolerance [2]. This work does not provide a detailed study on performance optimization and evaluation for issues specific to Web servers. Our contribution is to develop a resource management scheme for clustering Web servers with a master/slave (M/S) architecture and provide necessary scheduling optimization which considers the characteristics of Web workloads. The proposed techniques include separation of static and dynamic content processing, low-overhead remote CGI execution, and a reservation-based scheduler which considers both I/O and CPU utilization. Our experimental results show that the proposed optimization can lead to a 68% performance improvement. A detailed report on this work can be found in [6]. In evaluating the proposed techniques, we use the stretch factor [5], which is the ratio of response time with resourcesharing among requests to that without sharing. While there are other performance measurement metrics available, stretch factor is more suited for describing Web server performance because it represents a trade-off that allows short Web requests not to be slowed down too much by large jobs. This choice can meet user’s psychological expectation: in a system with highly variable task sizes, users may be willing to wait longer for large tasks to complete, but expect that small tasks should complete quickly [l].