Evgenia Smirni

New Results on the Performance Effects of Autocorrelated Flows in Systems (2009)

2 MAMSolver framework MAMSolver deals with the solution of M/G/1, GI/M/1, and QBD processes. Throughout this exposition, we refer to continuous time Markov processes. However MAMSolver provides...

Evgenia Smirni, Qi Zhang, Ningfang Mi, Alma Riska, Giuliano Casale

Temporal dependence within the workload of any computing or networking system has been widely recognized as a significant factor affecting performance. More specifically, burstiness, as a form of...

Anomaly? Application Change? or Workload Change? Towards Automated Detection of Application Performance Anomaly and Change ∗ (2009)

Ludmila Cherkasova, Kivanc Ozonat, Ningfang Mi, Julie Symons, Evgenia Smirni

Automated tools for understanding application behavior and its changes during the application life-cycle are essential for many performance analysis and debugging tasks. Application performance...

Bottlenecks and their Performance Implications inE-Commerce Systems (2009)

Qi Zhang, Alma Riska, Erik Riedel, Evgenia Smirni

Abstract. We present a detailed workload characterization of a multi-tiered sys-tem that hosts an e-commerce site. Using the TPC-W workload and via experimental measurements, we illustrate how...

Evaluating the Performability of Systems with Background Jobs ∗ (2008)

Qi Zhang, Alma Riska, Ningfang Mi, Erik Riedel, Evgenia Smirni

As most computer systems are expected to remain operational 24 hours a day, 7 days a week, they must complete maintenance work while in operation. This work is in addition to the regular tasks of the...

Acronyms (2008)

and GI/M/1-type processes; matrix analytic techniques; stochastic complementation). Summary and Conclusions We present an exact decomposition algorithm for the analysis of Markov chains with a...

conditions (2008)

Alma Riska, Wei Sun, Evgenia Smirni, Gianfranco Ciardo

effective balancing in clustered web servers under transient load

Abstract (2008)

Andreas Stathopoulos, Alma Riska, Evgenia Smirni, Zhili Hua

solution of M/G/1-type processes

class of (2008)

ETAQA Solutions for Infinite Markov Processes with Repetitive Structure (2008)

Extended abstract submitted for presentation at PMCCS-4 Projection: An efficient solution algorithm for a

Self-Adaptive Scheduler Parameterization via Online Simulation (2008)

Barry Lawson, Evgenia Smirni

High-end parallel systems present a tremendous research challenge on how to best allocate their resources to match dynamic workload characteristics and user habits that are often unique to each...

Categories and Subject Descriptors C.4 [Performance of Systems]: Design Studies General Terms Performance, Reliability (2008)

Ningfang Mi, Evgenia Smirni

As computer systems operate 24 hours a day, 7 days a week, it is becoming common to schedule maintenance jobs during idle times. These jobs are considered background jobs. Background jobs are used...

BRIDGING ETAQA AND RAMASWAMI’S FORMULA FOR THE SOLUTION OF M/G/1-TYPE PROCESSES ∗ (2008)

Andreas Stathopoulos, Alma Riska, Zhili Hua, Evgenia Smirni

For some time, the method of Ramaswami has been the established way to analyze M/G/1-type processes. The ETAQA method, proposed previously in [15], has offered a more efficient alternative for the...

Models of the Departure Process of a BMAP/MAP/1 queue 1 (2008)

Qi Zhang, Armin Heindl, Evgenia Smirni

Complex computer and communication systems are often modeled by queueing networks, with their arrival and/or service processes exhibiting correlations. In such systems, customers (or packets) may...

Interarrival Times Characterization and Fitting for Markovian Traffic Analysis (2008)

Casale, Giuliano, Zhang, Eddy Z., Smirni, Evgenia

We propose a traffic fitting algorithm for Markovian Arrival Processes (MAPs) that can capture statistics of any order of interarrival times. By studying real traffic traces, we show that matching...

Scheduling for performance and availability in systems with temporal dependent workloads (2008)

Ningfang Mi, Giuliano Casale, Evgenia Smirni

Temporal locality in workloads creates conditions in which a server, in order to remain available, should quickly process bursts of requests with large service requirements. In this paper, we show...

ETAQA-MG1: An Ecient Technique for the Analysis of M/G/1-type processes by Aggregation (2007)

Gianfranco Ciardo, Weizhen Mao, Alma Riska, Evgenia Smirni

We extend the ETAQA approach, initially proposed for the ecient numerical solution of a class of quasi birth-death processes, to the more complex case of M/G/1-type Markov processes. The new...

Cv (2007)

Evgenia Smirni

this paper appeared as ORNL technical report, ORNL/TM-12287. Also appeared as a book chapter in "Multiprocessor Performance Measurement and Evaluation ", pp 292-303, L.N. Bhuyan and X....

Projection: An efficient solution algorithm for a class of quasi-birth-death processes (2007)

Projection: An efficient solution algorithm for a (2007)

F49.43> 0 0 0 \Delta \Delta \Delta B (10) L (11) F 0 0 \Delta \Delta \Delta 0 B L F 0 \Delta \Delta \Delta 0 0 B L F \Delta \Delta \Delta . . . . . . . . . . . . . . . . . . 3 7 7 7 7 7 5 ; (1)...

Extended abstract submitted for presentation at PMCCS-4

class of (2007)

Exact analysis of a class of GI/G/1-type performability (2007)

Extended abstract submitted for presentation at PMCCS-4 Projection: An ecient solution algorithm for a

Algorithmic modications to the Jacobi-Davidson parallel eigensolver to dynamically balance external CPU (2007)

models

Richard Tran Mills, Andreas Stathopoulos, Evgenia Smirni

and memory load

Exact analysis of a class of GI/G/1-type performability models (2007)

Efficient Fitting of Long-Tailed Data Sets into Hyperexponential Distributions (2007)

this paper, we consider the solution of a class of CTMCs that show a GI/G/1-type repetitive structure, that is, they exhibit both M/G/1-type and GI/M/1-type patterns, and cannot be solved by existing...

Alma Riska, Vesselin Diev, Evgenia Smirni

We propose a new technique for fitting longtailed data sets into hyperexponential distributions. The approach partitions the data set in a divide and conquer fashion and uses the...

A Regression-Based Analytic Model for Dynamic Resource Provisioning of Multi-Tier Applications (2007)

Ludmila Cherkasova, Evgenia Smirni, Qi Zhang, Qi Zhang

three-tierapplications,

R-Capriccio: A Capacity Planning and Anomaly Detection Tool for Enterprise Services with Live (2007)

Qi Zhang, Lucy Cherkasova, Guy Matthews, Wayne Greene, Evgenia Smirni, Qi Zhang, ...

three-tierapplications, workload analysis, production environment, regression, transaction cost, capacity planning, performance modeling As the complexity of IT systems increases, performance...

Flexsplit: A workload-aware, adaptive load balancing strategy for media clusters (2006)

Qi Zhang, Ludmila Cherkasova, Evgenia Smirni

Abstract A number of technology and workload trends motivate us to consider a new request distribution and load balancing strategy for streaming media cluster. First, in emerging media workloads, a...

Load unbalancing to improve performance under autocorrelated traffic (2006)

Qi Zhang, Ningfang Mi, Alma Riska, Evgenia Smirni

Size-based policies have been shown to successfully balance load and improve performance in homogeneous cluster environments where a dispatcher assigns a job to a server strictly based on the job...

Evaluating the performability of systems with background jobs (2006)

Qi Zhang, Alma Riska, Ningfang Mi, Evgenia Smirni

As most computer systems are expected to remain operational 24 hours a day, 7 days a week, they must complete maintenance work while in operation. This work is in addition to the regular tasks of the...

Workload-aware load balancing for clustered web servers (2005)

Qi Zhang, Alma Riska, Wei Sun, Evgenia Smirni, Gianfranco Ciardo, Senior Member

Abstract—We focus on load balancing policies for homogeneous clustered Web servers that tune their parameters on-the-fly to adapt to changes in the arrival rates and service times of incoming...

Workload-aware load balancing for clustered web servers (2005)

Qi Zhang, Alma Riska, Wei Sun, Evgenia Smirni, Gianfranco Ciardo

Abstract We focus on load balancing policies for clustered web servers that tune their parameters on-the-fly to adapt to changes in the arrival rates and service times of incoming requests. The...

Characterizing the BMAP/MAP/1 departure process via the ETAQA truncation (2005)

Qi Zhang, Armin Heindl, Evgenia Smirni

We propose a family of finite approximations for the departure process of a BMAP/MAP/1 queue. The departure process approximations are derived via an exact aggregate solution technique (called ETAQA)...

Power-aware resource allocation in highend systems via online simulation (2005)

Barry Lawson, Evgenia Smirni

Traditionally, scheduling in high-end parallel systems focuses on how to minimize the average job waiting time and on how to maximize the overall system utilization. Despite the development of...

a class of M/G/1-type processes by aggregation (2005)

Gianfranco Ciardo, Weizhen Mao, Alma Riska, Evgenia Smirni

An efficient technique for the analysis of

Power-aware resource allocation in highend systems via online simulation (2005)

Barry Lawson, Evgenia Smirni

Traditionally, scheduling in high-end parallel systems focuses on how to minimize the average job waiting time and on how to maximize the overall system utilization. Despite the development of...

Characterizing the BMAP/MAP/1 departure process via the ETAQA truncation. Commun. Statist.-Stochastic Models (2005)

Qi Zhang, Armin Heindl, Evgenia Smirni

We propose a family of finite approximations for the departure process of a BMAP/MAP/1 queue. The departure process approximations are derived via an exact aggregate solution technique (called ETAQA)...

Workload-aware load balancing for clustered web servers (2005)

Qi Zhang, Alma Riska, Wei Sun, Evgenia Smirni, Gianfranco Ciardo

We focus on load balancing policies for homogeneous clustered web servers that tune their parameters on-the-fly to adapt to changes in the arrival rates and service times of incoming requests. The...

Workload-aware load balancing for clustered web servers (2005)

Qi Zhang, Alma Riska, Wei Sun, Evgenia Smirni, Gianfranco Ciardo

We focus on load balancing policies for clustered web servers that tune their parameters on-the-fly to adapt to changes in the arrival rates and service times of incoming requests. The proposed...

Workload-aware load balancing for clustered web servers (2005)

Qi Zhang, Alma Riska, Wei Sun, Evgenia Smirni, Gianfranco Ciardo, Senior Member

Abstract—We focus on load balancing policies for homogeneous clustered Web servers that tune their parameters on-the-fly to adapt to changes in the arrival rates and service times of incoming...

Etaqa-MG1: an efficient technique for the analysis of a class of M/G/1-type processes by aggregation (2004)

Gianfranco Ciardo, Weizhen Mao, Alma Riska, Evgenia Smirni

We extend the Etaqa approach, initially proposed for the efficient numerical solution of a class of quasi-birth–death processes, to a more complex class of M/G/1-type Markov processes where...

ETAQA Truncation Models for the MAP/MAP/1 Departure Process (2004)

Armin Heindl, Qi Zhang, Evgenia Smirni

We propose a family of finite approximations for the departure process of a MAP/MAP/1 queue. The departure process approximations are derived via an exact aggregate solution technique (called ETAQA)...

Bottlenecks and their performance implications in e-commerce systems (2004)

Qi Zhang, Alma Riska, Erik Riedel, Evgenia Smirni

Abstract. We present a detailed workload characterization of a multi-tiered system that hosts an e-commerce site. Using the TPC-W workload and via experimental measurements, we illustrate how...

Bottlenecks and their performance implications in e-commerce systems (2004)

Qi Zhang, Alma Riska, Erik Riedel, Evgenia Smirni

Abstract. We present a detailed workload characterization of a multi-tiered system that hosts an e-commerce site. Using the TPC-W workload and via experimental measurements, we illustrate how...

Profit-driven Service Differentiation in Transient (2003)

Environments Qi Zhang, Qi Zhang, Evgenia Smirni, Gianfranco Ciardo

We focus on service differentiation policies for Web servers that serve clients with different QoS requirements (i.e., different classes of service) in a constantly changing environment (i.e., an...

An EM-based technique for approximating long-tailed data sets with PH distributions (2003)

Alma Riska, Vesselin Diev, Evgenia Smirni

We propose a new technique for tting long-tailed data sets into phase-type (PH) distributions. This technique ts data sets with non-monotone densities into a mixture of Erlang and hyperexponential...

M/G/1-type Markov processes: A tutorial (2002)

M/G/1-type Markov processes: A tutorial (2002)

Abstract. M/G/1-type processes are commonly encountered when modeling modern complex computer and communication systems. In this tutorial, we present a detailed survey of existing solution methods...

Multiple-Queue Backfilling Scheduling with Priorities and Reservations for Parallel Systems (2002)

Barry G. Lawson, Evgenia Smirni

Abstract. We describe a new, non-FCFS policy to schedule parallel jobs on systems that may be part of a computational grid. Our algorithm continuously monitors the system (i.e., the intensity of...

ADAPTLOAD: effective balancing in clustered Web servers under transient load conditions (2002)

Alma Riska, Wei Sun, Evgenia Smirni, Gianfranco Ciardo

conditions

MAMSolver: A matrix analytic methods tool (2002)

Multiple-Queue Back (2002)

MAMsolver is a software tool for the solution of M/G/1type, GI/M/1-type, and QBD processes. The collection of solution algorithms implemented by MAMsolver are known as matrix-analytic methods and are...

Lling Scheduling With, Barry G. Lawson, Evgenia Smirni

We describe a new, non-FCFS policy to schedule parallel jobs on systems that may be part of a computational grid. Our algorithm continuously monitors the system (i.e., the intensity of incoming jobs...

M/G/1-type Markov Processes: A tutorial (2002)

Self-adapting Backfilling Scheduling for Parallel Systems (2002)

M/G/1-type processes are commonly encountered when modeling modern complex computer and communication systems. In this tutorial, we present a detailed survey of existing solution methods for...

Barry G. Lawson, Evgenia Smirni, Daniela Puiu

We focus on non-FCFS job scheduling policies for parallel systems that allow jobs to backfill, i.e., to move ahead in the queue, given that they do not delay certain previously submitted jobs....

Exact Aggregate Solutions for M/G/1-type Markov Processes (2002)

ADAPTLOAD: effective balancing in clustered web servers under transient load (2002)

We introduce a new methodology for the exact analysis of M/G/1-type Markov processes. The methodology uses basic, well-known results for Markov chains by exploiting the structure of the repetitive...

Conditions Alma Riska, Alma Riska, Wei Sun, Evgenia Smirni, Gianfranco Ciardo

We focus on adaptive policies for load balancing in clustered web servers, based on the size distribution of the requested documents. The proposed scheduling policy, ADAPTLOAD, adapts its balancing...

Models of Parallel Applications with Large Computation and I/O Requirements (2002)

Emilia Rosti, Giuseppe Serazzi, Evgenia Smirni, Ieee Computer Society, Ieee Computer Society, Mark S. Squillante, ...

this paper, we present a formal model of the behavior of CPU and I/O interactions in scientific applications, from which we derive various formulas that characterize application performance. Our...

EQUILOAD: a load balancing policy for clustered Web servers. Performance Evaluation (2001)

EQUILOAD: a load balancing policy for clustered Web servers. Performance Evaluation (2001)

We present a new strategy for the allocation of requests in clustered web servers, based on the size distribution of the requested documents. This strategy, EquiLoad, manages to achieve a balanced...

EQUILOAD: a load balancing policy for clustered Web servers. Performance Evaluation (2001)

We present a new strategy for the allocation of requests in clustered web servers, based on the size distribution of the requested documents. This strategy, EquiLoad, manages to achieve a balanced...

Algorithmic modifications to the Jacobi-Davidson parallel eigensolver to dynamically balance external CPU and memory load (2001)

We present a new strategy for the allocation of requests in clustered web servers, based on the size distribution of the requested documents. This strategy, EquiLoad, manages to achieve a balanced...

Richard Tran Mills, Andreas Stathopoulos, Evgenia Smirni

Clusters of workstations (COWs) and SMPs have become popular and cost eective means of solving scientic problems. Because such environments may be heterogenous and/ or time shared, dynamic load...

EQUILOAD: a load balancing policy for clustered Web servers. Performance Evaluation (2001)

Analytic modeling of load balancing policies for tasks with heavy-tailed distributions (2000)

We present a new strategy for the allocation of requests in clustered web servers, based on the size distribution of the requested documents. This strategy, EquiLoad, manages to achieve a balanced...

Andrew S. Miner, Gianfranco Ciardo, Evgenia Smirni, Andreas Stathopoulos, Virginia Torczon, Alex Pothen

We present an analytic technique for modeling load balancing policies on a cluster of servers conditioned on the fact that the service times of arriving tasks are drawn from heavy tail distributions....

APPROVAL SHEET (2000)

ii To my mother, who bought my first computer iii

Analytic modeling of load balancing policies for tasks with heavy-tailed distributions (2000)

Analytic modeling of load balancing policies for tasks with heavy-tailed distributions (2000)

Abstract We present an analytic technique for modeling load balancing policies on a cluster of servers conditioned on the fact that the arriving tasks are drawn from heavy tail distributions. We...

Analytic modeling of load balancing policies for tasks with heavy-tailed distributions (2000)

We present an analytic technique for modeling load balancing policies on a cluster of servers conditioned on the fact that the service times of arriving tasks are drawn from heavy tail distributions....

An aggregation-based solution method for M/G/1-type processes (1999)

We present an analytic technique for modeling load balancing policies on a cluster of servers conditioned on the fact that the service times of arriving tasks are drawn from heavy tail distributions....

ETAQA: an efficient technique for the analysis of QBD-processes by aggregation (1999)

Abstract. We extend the ETAQA approach, initially proposed for the efficient numerical solution of a class of quasi birth-death processes, to the more complex case of M/G/1-type Markov processes. The...

An aggregation-based solution method for M/G/1-type processes (1999)

An aggregation-based solution method for M/G/1-type processes (1999)

Abstract. We extend the ETAQA approach, initially proposed for the ecient numerical solution of a class of quasi birth-death processes, to the more complex case of M/G/1-type Markov processes. The...

Adaptive CPU Scheduling Policies for Mixed Multimedia and Best-effort Workloads (1999)

. We extend the ETAQA approach, initially proposed for the efficient numerical solution of a class of quasi birth-death processes, to the more complex case of M/G/1-type Markov processes. The...

Melissa A. Rau, Evgenia Smirni

As multimedia applications with real-time constraints rapidly invade today's desktops, it becomes increasingly important for the operating system to provide robust resource allocation mechanisms...

An aggregation-based solution method for M/G/1-type processes (1999)

Performance Evaluation of Parallel Systems (1999)

We extend the ETAQA approach, initially proposed for the efficient numerical solution of a class of quasi birth-death processes, to the more complex case of M/G/1-type Markov processes. The proposed...

Paolo Cremonesi, Emilia Rosti, Giuseppe Serazzi, Evgenia Smirni

In this paper performance evaluation methodologies that have been applied in the analysis of parallel systems are reviewed together with the specific performance metrics. We concentrate on a few...

Performance Evaluation of Parallel Systems (1999)

Paolo Cremonesi, Emilia Rosti, Evgenia Smirni

In this paper performance evaluation methodologies that have been applied in the analysis of parallel systems are reviewed together with the specific performance metrics. We concentrate on a few...

ETAQA: an efficient technique for the analysis of QBD-processes by aggregation. Performance Evaluation (1999)

The Impact of I/O on Program Behavior and Parallel Scheduling (1998)

In this paper we present ETAQA, an Efficient Technique for the Analysis of QBD-processes by Aggregation. We concentrate on processes satisfying a particular repetitive structure that frequently...

Emilia Rosti, Giuseppe Serazzi, Evgenia Smirni, Mark S. Squillante

In this paper we systematically examine various performance issues involved in the coordinated allocation of processor and disk resources in large-scale parallel computer systems. Models are...

Lessons from Characterizing Input/Output Bahavior of Parallel Scientific Applications (1998)

Corresponding Evgenia Smirni, Evgenia Smirni, Daniel Reed

Because both processor and interprocessor communication hardware is evolving rapidly with only moderate improvements to file system performance in parallel systems, it is becoming increasingly...

A Workload Characterization Study of I/O Intensive Parallel Programs (1998)

Emilia Rosti, Giuseppe Serazzi, Evgenia Smirni, Mark S. Squillante

Characterizing the I/O requirements of parallel applications that manipulate huge amounts of data, such as scientific codes, is critical to the achievement of good application performance and to the...

Processor Saving Scheduling Policies for Multiprocessor Systems (1998)

Emilia Rosti, Evgenia Smirni, Ieee Computer Society, Kenneth C. Sevcik

In this paper, processor scheduling policies that "save" processors are introduced and studied. In a multiprogrammed parallel system, a "processor saving" scheduling policy...

Performance Modeling of a Parallel I/O System: An Application Driven Approach (1997)

Evgenia Smirni, Christopher L. Elford, Daniel A. Reed, Andrew A. Chien

The broadening disparity between the performance of I/O devices and the performance of processors and communication links on parallel platforms is a major obstacle to achieving high performance in...

Workload Characterization of Input/Output Intensive Parallel Applications (1997)

Evgenia Smirni, Daniel A. Reed

. The broadening disparity in the performance of input/output (I/O) devices and the performance of processors and communication links on parallel systems is a major obstacle to achieving high...

Workload Characterization of Input/Output Intensive Parallel Applications (1997)

Corresponding Evgenia Smirni, Evgenia Smirni, Daniel A. Reed

Because both processor and interprocessor communication hardware is evolving rapidly with only moderate improvements to file system performance in parallel systems, it is becoming increasingly...

Algorithmic Influences on I/O Access Patterns and Parallel File System Performance (1997)

Evgenia Smirni, C. E. Elford, A. J. Lavery, A. A. Chien

For many scalable parallel applications, the input /output (I/O) barrier rivals or exceeds that of computation and interprocessor communication. Consequently, scalable parallel secondary and tertiary...

I/O Requirements of Scientific Applications: An Evolutionary View (1996)

Evgenia Smirni, Ruth A. Aydt, Andrew A. Chien, Daniel A. Reed

The modest I/O configurations and file system limitations of many current high-performance systems preclude solution of problems with large I/O needs. I/O hardware and file system parallelism is the...

The Next Frontier: Interactive and Closed Loop Performance Steering (1996)

Daniel A. Reed, Christopher L. Elford, Tara M. Madhyastha, Evgenia Smirni, Stephen E. Lamm

Software for a growing number of problem domains has complex, time varying behavior and unpredictable resource demands (e.g., WWW servers and parallel input /output systems). While current...

I/O Requirements of Scientific Applications: An Evolutionary View (1996)

Evgenia Smirni, Ruth A. Aydt, Andrew A. Chien, Daniel A. Reed

The modest I/O configurations and file system limitations of many current high-performance systems preclude solution of problems with large I/O needs. I/O hardware and file system parallelism is the...

I/O, Performance Analysis, and Performance Data Immersion (1996)

Daniel A. Reed, Christopher L. Elford, Tara Madhyastha, Will H. Scullin, Ruth A. Aydt, Evgenia Smirni

A large and important class of national challenge applications are irregular, with complex, data dependent execution behavior, and dynamic, with time varying resource demands. We believe the solution...

Performance Visualization: 2-D, 3-D, and Beyond (1996)

Daniel A. Reed, Matthew J. Gardner, Evgenia Smirni

During the past ten years, performance data visualization techniques have evolved from static, twodimensional graphics to dynamic graphics and immersive virtual environments. We sketch the domain of...

Processor allocation and thread placement policies in parallel multiprocessor systems /--by Evgenia Smirni. (1995)

Smirni, Evgenia.

Thesis (Ph. D. in Computer Science)--Vanderbilt University, 1995.

PerPreT - A Performance Prediction Tool for Massively Parallel Systems (1995)

Jürgen Brehm, Manish Madhukar, Evgenia Smirni, Larry Dowdy

Today's massively parallel machines are typically message passing systems consisting of hundreds or thousands of processors. Implementing parallel applications efficiently in this environment is...

Efficient fitting of long-tailed data sets into PH distributions. Performance Evaluation, 2003 (to appear (1975)

Alma Riska, Vesselin Diev, Evgenia Smirni

We propose a new technique for fitting long-tailed data sets into phase-type (PH) distributions. This technique fits data sets with non-monotone densities into a mixture of Erlang and...

ETAQA: An Efficient Technique for the Analysis of QBD-processes by Aggregation