Yong Meng Teo

Dynamic Resource Pricing on Federated Clouds

Current large distributed systems allow users to share and trade resources. In cloud computing, users purchase different types of resources from one or more resource providers using a fixed pricing scheme. Federated clouds, a topic of recent interest, allows different cloud providers to share resources for increased scalability and reliability. However, users and providers of cloud resources are rational and maximize their own interest when consuming and contributing shared resources. In this paper, we present a dyanmic pricing scheme suitable for rational users requests containing multiple resource types. Using simulations, we compare the efficiency of our proposed strategy-proof dynamic scheme with fixed pricing, and show that user welfare and the percentage of successful requests is increased by using dynamic pricing.

Sensor grid: integration of wireless sensor networks and the grid

Wireless sensor networks have emerged as an exciting technology for a wide range of important applications that acquire and process information from the physical world. Grid computing has evolved as a standards-based approach for coordinated resource sharing. Sensor grids combine these two promising technologies by extending the grid computing paradigm to the sharing of sensor resources in wireless sensor networks. There are several issues and challenges in the design of sensor grids. In this paper, we propose a sensor grid architecture, called the scalable proxy-based architecture for sensor grid (SPRING), to address these design issues. We also developed a sensor grid testbed to study the design issues of sensor grids and to improve our sensor grid architecture design

Comparison of Load Balancing Strategies on Cluster-based Web Servers

This paper focuses on an experimental analysis of the perfor mance and scalability of cluster-based web servers. We carry out the comparative studies using two experimental platforms, namely, a hardware testbed consisting of sixteen PCs, and a trace-driven discrete-event simulator. Dispatcher and web server service times used in the simulator are determined by carrying out a set of experiments on the testbed. The simulator is validated against stochastic queuing models and the testbed. Experiments on the testbed are limited by the hardware configu ration, but our complementary approach allows us to carry out scalability studies on the validated simulator. The three dis patcher-based scheduling algorithms analyzed are: round robin scheduling, least connected based scheduling, and least loaded based scheduling. The least loaded algorithm is used as the baseline (upper performance bound) in our analysis and the performance metrics include average waiting time, average re sponse time, and average web server utilization. A synthetic trace generated by the workload generator called SURGE, and a public-domain France Football World Cup 1998 trace are used. We observe that the round robin algorithm performs much worse in comparison with the other two algorithms for low to medium workload. However, as the request arrival rate increases, the performance of the three algorithms converge with the least con nected algorithm approaching the baseline algorithm at a much faster rate than the round robin. The least connected algorithm performs well for medium to high workload. At very low load, the average waiting time is two to six times higher than the baseline algorithm but the absolute value between these two waiting times is very small.

Speculative parallel simulation with an adaptive throttle scheme

Excessive rollback recoveries due to overoptimistic event execution in Time Warp simulators often degrade their runtime performance. This paper presents a two-sided throttling scheme to dynamically adjust the event execution speed of Time Warp simulators. The proposed throttle is based on a new concept called global progress window, which allows the individual simulation process to be positioned on a global time scale, thereby to accelerate or suspend their event execution. As each simulation process can be throttled to a steady state, excessive rollback recoveries due to causality errors can be avoided. To quantify the effect of rollbacks and for purpose of comparing different Time Warp implementations, we propose two new measures called RPE (number of Rollback events Per committed Event), and E (relative Effectiveness in reducing rollback overhead). Our implementation results show that the proposed throttle effectively regulates the proceeding of each simulation process, resulting in a significant reduction in rollback thrashing and elapsed time.

CODES: An Integrated Approach to Composable Modeling and Simulation

In component-based simulation, models developed in different locations and for specific purposes can be selected and assembled in various combinations to meet diverse user requirements. This paper proposes CODES (COmposable Discrete-Event scalable Simulation), an approach to component-based modeling and simulation that supports model reuse across multiple application domains. A simulation component is viewed by the modeller as a black box with an in- and/or out-channel. The attributes and behavior of the component abstracted as a meta-component are described using COML (COmponent Markup Language), a markup language we propose for representing simulation components. The integrated approach, supported by a proposed COSMO (COmponent-oriented Simulation and Modeling Ontology) ontology, consists of four main steps. Component discovery returns a set of syntactically valid model components. Syntactic composability is determined by our proposed EBNF syntactic composition rules. Validation of semantic composability is performed using our proposed data and behavior alignment algorithms. The semantically valid simulation component is subsequently stored in a model repository for reuse. As proof of concept, we discuss a prototype implementation of the CODES framework using queueing system as an application domain example.

On Economic and Computational-Efficient Resource Pricing in Large Distributed Systems

There is growing interest in large-scale systems where globally distributed and commoditized resources can be shared and traded, such as peer-to-peer networks, grids, and cloud computing. Users of these systems are rational and maximize their own interest when consuming and contributing shared resources, even if by doing so they affect the overall efficiency of the system. To manage rational users, resource pricing and allocation can provide the necessary incentives for users to behave such that the overall efficiency can be maximized. In this paper, we propose a dynamic pricing mechanism for the allocation of shared resources, and evaluate its performance. In contrast with several existing trading models, our scheme is designed to allocate a request with multiple resource types, such that the user does not have to aggregate different resource types manually. We formally prove the economic properties of our pricing scheme using the mechanism design framework. We perform both theoretical and simulation analysis to evaluate the economic and computational efficiency of the allocation and the scalability of the mechanism. Our simulations are validated against a prototype implementation on PlanetLab.

On understanding the energy consumption of ARM-based multicore servers

There is growing interest to replace traditional servers with low-power multicore systems such as ARM Cortex-A9. However, such systems are typically provisioned for mobile applications that have lower memory and I/O requirements than server application. Thus, the impact and extent of the imbalance between application and system resources in exploiting energy efficient execution of server workloads is unclear. This paper proposes a trace-driven analytical model for understanding the energy performance of server workloads on ARM Cortex-A9 multicore systems. Key to our approach is the modeling of the degrees of CPU core, memory and I/O resource overlap, and in estimating the number of cores and clock frequency that optimizes energy performance without compromising execution time. Since energy usage is the product of utilized power and execution time, the model first estimates the execution time of a program. CPU time, which accounts for both cores and memory response time, is modeled as an M/G/1 queuing system. Workload characterization of high performance computing, web hosting and financial computing applications shows that bursty memory traffic fits a Pareto distribution, and non-bursty memory traffic is exponentially distributed. Our analysis using these server workloads reveals that not all server workloads might benefit from higher number of cores or clock frequencies. Applying our model, we predict the configurations that increase energy efficiency by 10% without turning off cores, and up to one third with shutting down unutilized cores. For memory-bounded programs, we show that the limited memory bandwidth might increase both execution time and energy usage, to the point where energy cost might be higher than on a typical x64 multicore system. Lastly, we show that increasing memory and I/O bandwidth can improve both the execution time and the energy usage of server workloads on ARM Cortex-A9 systems.

An adaptive stabilization framework for distributed hash tables

Distributed hash tables (DHT) algorithms obtain good lookup performance bounds by using deterministic rules to organize peer nodes into an overlay network. To preserve the invariants of the overlay network, DHTs use stabilization procedures that reorganize the topology graph when participating nodes join or fail. Most DHTs use periodic stabilization, in which peers perform stabilization at fixed intervals of time, disregarding the rate of change in overlay topology; this may lead to poor performance and large stabilization-induced communication overhead. We propose a novel adaptive stabilization framework that takes into consideration the continuous evolution in network conditions. Each peer collects statistical data about the network and dynamically adjusts its stabilization rate based on the analysis of the data. The objective of our scheme is to maintain nominal network performance and to minimize the communication overhead of stabilization.

On Syntactic Composability and Model Reuse

Composability, the capability to select and assemble off-the-shelf model components in various combinations to satisfy user requirements, is an appealing approach in reducing the time and costs of developing complex simulation. This paper discusses CODES, a hierarchical component framework to support component-based modeling and simulation. We propose the use of EBNF based grammars to specify syntactic composability rules with the aims of achieving syntax consistency for model components to operate together. EBNF production strings associated with each composed models are transformed into a unique identifier to support distributed DHT-based model discovery. The hierarchical design supports the sharing and reuse of models and model components across application domains, and facilitates the verification of composed models. We present a prototype of the framework implemented using the scalable simulation framework, and illustrate this approach by modeling a grid computing system

A performance study of big data on small nodes

The continuous increase in volume, variety and velocity of Big Data exposes datacenter resource scaling to an energy utilization problem. Traditionally, datacenters employ x86-64 (big) server nodes with power usage of tens to hundreds of Watts. But lately, low-power (small) systems originally developed for mobile devices have seen significant improvements in performance. These improvements could lead to the adoption of such small systems in servers, as announced by major industry players. In this context, we systematically conduct a performance study of Big Data execution on small nodes in comparison with traditional big nodes, and present insights that would be useful for future development. We run Hadoop MapReduce, MySQL and in-memory Shark workloads on clusters of ARM big. LITTLE boards and Intel Xeon server systems. We evaluate execution time, energy usage and total cost of running the workloads on self-hosted ARM and Xeon nodes. Our study shows that there is no one size fits all rule for judging the efficiency of executing Big Data workloads on small and big nodes. But small memory size, low memory and I/O bandwidths, and software immaturity concur in canceling the lower-power advantage of ARM servers. We show that I/O-intensive MapReduce workloads are more energy-efficient to run on Xeon nodes. In contrast, database query processing is always more energy-efficient on ARM servers, at the cost of slightly lower throughput. With minor software modifications, CPU-intensive MapReduce workloads are almost four times cheaper to execute on ARM servers.