Prashant Pradhan

Provisioning servers in the application tier for e-commerce systems

Server providers that support e-commerce applications as a service to multiple e-commerce websites traditionally use a tiered server architecture. This architecture includes an application tier to process requests that require dynamically generated content. How this tier is provisioned can significantly impact a providers profit margin. We study methods to provision servers in the application serving tier to increase a server providers profits. First, we examine actual traces of request arrivals to the application tier of e-commerce sites, and show that the arrival process is effectively Poisson. Next, we construct an optimization problem in the context of a set of application servers modeled as M/G/l/PS queueing systems, and derive three simple methods to approximate the allocation that maximizes profits. Simulation results demonstrate that our approximation methods achieve profits that are close to optimal and are significantly higher than those achieved via simple heuristics.

An observation-based approach towards self-managing Web servers

The Web server architectures that provide performance isolation, service differentiation, and QoS guarantees rely on external administrators to set the right parameter values for the desired performance. Due to the complexity of handling varying workloads and bottleneck resources, configuring such parameters optimally becomes a challenge. In this paper we describe an observation-based approach for self-managing Web servers that can adapt to changing workloads while maintaining the QoS requirements of different classes. In this approach, the system state is monitored continuously and parameter values of various system resources-primarily the accept queue and the CPU-are adjusted to maintain the system-wide QoS goals. We implement our techniques using the Apache Web server and the Linux operating system. We first demonstrate the need to manage different resources in the system depending on the workload characteristics. We then experimentally demonstrate that our observation-based system can adapt to workload changes by dynamically adjusting the resource shares in order to maintain the QoS goals.

Spatio-Temporal Patterns for Problem Determination in IT Services

Problem determination in a large and dynamic IT service is a challenging task. In this paper we propose a framework for problem determination based on monitoring the event streams generated by the different components of an IT service. We give a generic representation of a problem through spatial-temporal patterns, which is a graph where the vertices capture the location and the time of the matching events, and the edges represent the spatio-temporal conditions between two matching events. The spatial conditions are based on the underlying system topology graph, and the temporal conditions are based on event timestamps.A practical implementation of the above framework will require fast algorithms for detecting patterns. We present efficient algorithms when the pattern graph is a range and a tree, which are then used as building blocks for a hierarchical heuristic for detecting general patterns. Finally, we show that our algorithms perform well in practice through extensive numerical simulations.

Efficient and Faithful Performance Modeling for Network-Processor-Based System Designs

Publisher Summary This chapter presents the design, implementation, and initial performance evaluation of a hybrid performance modeling framework called Countach, which faithfully and efficiently captures application behavior while predicting system performance. After a small manual step of replacing hardware interactions in the system software with simulated events, a set of automated postprocessing steps converts an application to a processor model that captures memory performance and enforces tight time synchronization with the simulator. Multiple processors or hardware execution contexts can be modeled in this framework. Actual workload generators can be used unmodified to drive the performance analysis. Countach models various kinds of network servers that utilize network-processor-based network interfaces. It is a very useful tool, because it allows users to experiment with the entire spectrum of design choices, including hardware design, a rich variety of operating system optimizations, and choices of function split between the host and the network processors.