ALMA RISKA

548 Juniper Ct	757-561-4938
Mars, PA, 16046	alma.riska@gmail.com
	http://www.cs.wm.edu/~riska

RESEARCH INTERESTS

 

System optimization, fault tolerant computing, resource management policies, data mining, analysis, and modeling, data-driven system evaluation and design, performance evaluation, reliability modeling, data availability and consistency, power-aware computing, stochastic processes.

EDUCATION

PhD in Computer Science	December 2002
College of William & Mary, Williamsburg, VA

Advisor:	Evgenia Smirni
Dissertation Title:	``Aggregate Matrix-Analytic techniques and their applications''

Diploma in Computer Science (5-year curriculum)	January 1996
University of Tirana, Tirana, Albania

Advisor:	Jochen W. Muench
Thesis Title:	``Traffic measurement on FDDI networks''

PROFESSIONAL EXPERIENCE

Research Staff Member	October 2002 - Present
Seagate Research, Pittsburgh, PA.

• System Reliability:
  • In-depth evaluation of a significant set of SMART and page logs collected from approximately half a million drives in the field over the course of three years (a total of 10 TB of data) and during the Field Return Incoming Test (FRIT) with the goal of characterizing the transient and non-transient faulty drive behavior in the field. Such characterization enables the evaluation of the failure prediction capabilities in the SMART logs and is the basis of new failure prediction algorithms that are devised to achieve high failure prediction rates while keeping the false positives low.
  • Modeling and evaluation of storage system-level benefits when predicting and identifying partial disk drive failures. Failure data from the Failure Analysis labs and the Field Return Incoming Test were used to quantify the improvement of data reliability in storage systems that are aware of partial failures.
  • Data-driven disk drive failure analysis using the page logs collected during the Field Return Incoming Test (FRIT) for the enterprise-level drives.
  • Developed a data-driven framework to evaluate the feasibility of Idle Read After Write (IRAW) [16], which detects and recovers data corruption.
  • Evaluated the benefits of pro-active techniques for detection and correction of latent sector errors [17].
  • Developed the set of data mining and analysis tools that enabled the above data-driven evaluation.
• System Design and Evaluation:
  • In-depth evaluation of the schedulability of not instantaneously preemptable background jobs in storage systems. The evaluation consisted of analytic models, trace-driven simulations, and measurements on prototypes. The analysis led to new algorithms that manage the available idleness in a system such that the background work completes without violating foreground performance targets [P1], [1],[12],[18],[23].
  • Measurement-based evaluation of the block-level optimization through the IO hierarchy, including the file system, the device driver, and the disk drive itself, with the goal to identify and quantify the collective effect the optimization efforts in various levels of the IO hierarchy have on the user perceived performance [19].
  • Disk-level workload characterization driven by:
    • traces measured in enterprise, desktop, and consumer electronic storage devices [20], [24],
    • data extracted from the page logs collected during the Field Return Incoming Test (FRIT) from the returned enterprise-level drives [14],
    • data extracted from SMART logs collected periodically from drives in the field [13].
    The analysis identified the salient characteristics of storage workloads, the commonalities, and differences across different computing environments.
    • Workload propagation through all tiers of a multi-tiered system focusing on the demands of the e-commerce applications from the underlying storage subsystem. The evaluation identified the main characteristics of the traffic flows that propagate through the tiers with correlation being the prominent one. This analysis was used to derive scheduling and resource management policies that are aware of these characteristics [3], [15], [26], [28].
    • Adaptive storage virtualization via a feedback control loop based on the monitored storage system latency and throughput [5], [25], [27].
    • Characterization and management of overload at various levels of the IO hierarchy of a computer system. The analysis was used to devise heuristic-based disk scheduling algorithms for efficient and adaptive operation of disk drives during overload conditions [P2], [P3], [29], [30], [31].
  • Collaboration and Mentoring
    • Initiated and maintained collaboration with several academic partners.
    • Mentored and guided interns that contributed in many of the above efforts.

  Research Assistant	January 1999 - October 2002
  Computer Science Department, The College of William and Mary, Williamsburg, VA.

  • Performance evaluation of clustered Web servers using analytic modeling, simulation, and workload prediction and characterization. New load balancing policies that sustain the transient overload of clustered Web servers were proposed [2], [6], [11], [21], [22], [34], [40].
  • Software development (C++) for analytic methods on the numerical solution of Markov processes with repetitive structure. Implemented the tool MAMSolver [38] and successfully used it to analyze the performance of clustered Web servers.
  • Efficient aggregation-based matrix-analytic methods for the solution of M/G/1 and GI/G/1-type processes. The proposed methodology, called {\Etaqa}, was used to model clustered Web servers and other complex computer systems [4], [8], [9], [32], [36], [41], [44].
  • Computationally efficient heuristics for fitting data sets that exhibit heavy tail in their empirical distribution into tractable distributions such as the hyperexponential and hypoexponential ones [10], [33], [37].

  Internship	Summer 2001
  IBM T.J. Watson Research Center, Hawthorne, NY.	Mentor: Mark S. Squillante.

  Analytic modeling of Web servers via workload characterization and the matrix-analytic methodology. A new fitting technique was devised, that parameterizes the workload and describes analytically Web server behavior. Simulations of Web servers using commercial Web server traces validate the accuracy of the new technique and of the overall approach [35].

  Internship	Summer 2000
  HP Research Laboratories Palo Alto, CA.	Mentor: Dejan Milojicic.

  Modeling, profiling, and simulation of the memory hierarchy of an IA-64 architecture, by taking into consideration its interleaving, read-write sequences, and soft errors, with the goal to investigate the availability of such systems [7], [39].

  Teaching Assistant	August 1998 - January 1999
  Computer Science Department, The College of William and Mary, Williamsburg, VA.

  Courses ``Introduction to C++'', ``Concepts in Computer Science''.

  Network Specialist	February 1995 - August 1998
  Computer Center, University of Siegen, Germany.

  Installation, configuration and measurement of the performance of the ATM network in the University (Cisco devices). Configuration of the Netware 4.0 in clustered Windows workstations.

  Database Software Engineer	October 1993 - February 1995
  International Computer Company, Tirana, Albania.

  Developed the database system for the administration of the Immovable Properties in Albania (a joint project with the US Department of Treasure and the University of Wisconsin at Madison). Developed a database system for real estate companies in Tirana.

  CONFERENCE PROGRAM COMMITTEES

  • 2010:
    • The USENIX Conference on File and Storage Technologies (FAST).
  • 2009:
    • The IEEE International Symposium on Modeling, Analysis, and Simulation of Computer and Telecommunication Systems (MASCOTS).
    • The Workshop on Run-time Models for Self-Managing Systems and Applications (ROSSA).
  • 2008:
    • The USENIX Conference on File and Storage Technologies (FAST).
    • The ACM SIGMETRICS Conference on Measurement and Modeling of Computer Systems (SIGMETRICS).
  • 2007
    • The International Conference on Quantitative Evaluation of Systems (QEST).
    • The IEEE International Symposium on Workload Characterization (IISWC).
  • 2006:
    • The ACM SIGMETRICS Conference on Measurement and Modeling of Computer Systems (SIGMETRICS).
    • The International Conference on Quantitative Evaluation of Systems (QEST).
    • The International Conference on Parallel and Distributed Systems (ICPADS).
  • 2005:
    • The International Conference on Quantitative Evaluation of Systems (QEST).
    • Co-chair of the International Workshop on Performability Modeling of Computer and Communication Systems (PMCCS).

  JOURNAL PUBLICATIONS

  [1]

  N. Mi, A. Riska, Q. Zhang, E. Smirni, E. Riedel, ``Efficient Management of Idleness in Storage Systems'', to appear in ACM Transactions on Storage.

  [2]

  Q. Zhang, N. Mi, A. Riska, E. Smirni, ``Performance-Guided Load (Un)balancing under Autocorrelated Flows'', IEEE Transactions on Parallel and Distributed Systems, Vol. 19, No. 5, pages 652-665, 2008.

  [3]

  N. Mi, Q. Zhang, A. Riska, E. Smirni, E. Riedel, ``Performance impacts of autocorrelated flows in multi-tiered systems'', Elsevier Performance Evaluation Journal, Vol. 64, No. 9-12, pages 1082-1101, 2007.

  [4]

  A. Riska, E. Smirni, ``ETAQA Solutions for Infinite Markov Processes with Repetitive Structure'', INFORMS Journal on Computing, Vol. 19, No. 2, pages 215-228, 2007.

  [5]

  J. Zhang, A. Sivasubramaniam, Q. Wang, A. Riska, E. Riedel, ``Storage Performance Virtualization via Throughput and Latency Control'', ACM Transactions on Storage, Vol. 2, No. 3, pages 283-308, 2006.

  [6]

  Q. Zhang, A. Riska, W. Sun, E. Smirni, G. Ciardo, ``Workload-aware load balancing for clustered Web servers'', IEEE Transactions on Parallel and Distributed Systems, Vol 16, No 3, pages 219-233, 2005.

  [7]

  A. Messer, P. Bernadat, G. Fu, D. Chen, Z. Dimitrijevic, D. Lie, D. Mannaru, A. Riska, D. Milojicic, ``Susceptibility of Commodity Systems and Software to Memory Soft Errors", IEEE Transactions on Computers, Vol 53, No 12, pages 1557-1568, 2004.

  [8]

  G. Ciardo, W. Mao, A. Riska, E. Smirni, ``ETAQA-MG1: an efficient technique for the analysis of M/G/1-type processes by aggregation'', Elsevier Performance Evaluation Journal, Vol. 57(3), pages 235-260, 2004.

  [9]

  A. Riska, E. Smirni, G. Ciardo, ``Exact analysis of a class of GI/G/1-type performability models'', IEEE Transactions on Reliability, Vol. 53, No 2, pages 238-249, 2004.

  [10]

  A. Riska, V. Diev, E. Smirni, ``Efficient fitting of long-tailed data sets into Phase-type distributions'', Elsevier Performance Evaluation Journal, Vol. 55, pages 147-164, 2004.

  [11]

  G. Ciardo, A. Riska, E. Smirni, ``EQUILOAD: a load balancing policy for clustered Web servers'', Elsevier Performance Evaluation Journal, Vol. 46, pages 101-124, 2001.

  REFEREED CONFERENCE AND WORKSHOP PUBLICATIONS

  [12]

  N. Mi, A. Riska, X. Li, E. Smirni, E. Riedel, ``Restraint Utilization of Idleness for Transparent Scheduling of Background Tasks'', to appear in Proceedings of the ACM SIGMETRICS Conference, Seattle, WA, June 2009.

  [13]

  A. Riska, E. Riedel, ``Evaluation of Disk-level Workloads at Different Time Scales'', to appear in Proceedings of the ACM SIGMETRICS Conference, (as an extended abstrat) Seattle, WA, June 2009.

  [14]

  A. Riska, E. Riedel, ``Disk Drive Logs Collected during the Field Return Incoming Test'', in Proceedings of the USENIX Workshop on Analysis of System Logs (WASL), co-located with OSDI, San Diego, CA, December 2008.

  [15]

  M. Zhang, A. Riska, E. Riedel, ``Characterization of the Storage Subsystem Workload under E-commerce Applications'', in Proceedings of the 5th International Conference on Quantitative Evaluation of Systems (QEST), pages 385-394, San Malo, France, September 2008.

  [16]

  A. Riska, E. Riedel, ``Idle Read After Write - IRAW'', in Proceedings of the Annual USENIX Technical Conference, pages 43-56, Boston, MA, June 2008.

  [17]

  N. Mi, A. Riska, E. Smirni, E. Riedel, ``Enhancing Data Availability through Background Activities'', in Proceedings of the Symposium on Dependable Systems and Networks (DSN), pages 492-501, Anchorage, AK, June 2008

  [18]

  N. Mi, A. Riska, Q. Zhang, E. Smirni, E. Riedel, ``Efficient management of idleness in systems'', in Proceedings of the ACM SIGMETRICS, extended abstract, pages 371-372, June 2007.

  [19]

  A. Riska, J. Larkby-Lahet, E. Riedel, ``Evaluating Block-level Optimization Through the IO Path'', in Proceedings of the Annual USENIX Technical Conference, pages 247-260, June 2007.

  [20]

  A. Riska, E. Riedel, ``Long-range dependence at the disk drive level'', in Proceedings of the 3rd International Conference on the Quantitative Evaluation of Systems (QEST), pages 41-50, Riverside, CA, September 2006.

  [21]

  N. Mi, Q. Zhang, A. Riska, E. Smirni, ``Load Balancing for Performance Differentiation in Dual-Priority Clustered Servers'', in Proceedings of the 3rd International Conference on Quantitative Evaluation of Systems (QEST), pages 385-394, Riverside, CA, September 2006.

  [22]

  Q. Zhang, N. Mi, A. Riska, E. Smirni, ``Load Unbalancing to Improve Performance under Autocorrelated Traffic'', in Proceedings of the International Conference on Distributed Computer Systems (ICDCS), pages 20-27, Lisboa, Portugal, July 2006.

  [23]

  Q. Zhang, A. Riska, E. Riedel, N. Mi, E. Smirni, ``Evaluating performability of systems with background jobs'', in Proceedings of the Symposium on Dependable Systems and Networks (DSN), pages 495-505, Philadelphia, June 2006. A shorter version presented at the International Workshop on Performability Modeling of Computer and Communication Systems (PMCCS), Torino, Italy, 2005.

  [24]

  A. Riska, E. Riedel, ``Disk Drive Level Workload Characterization'', in Proceedings of the Annual USENIX Technical Conference, pages 97-103, Boston, MA, June 2006.

  [25]

  J. Zhang, A. Riska, A. Sivasubramaniam, Q. Wang, E. Riedel, ``Storage Performance virtualization via Throughput and Latency Control'', in Proceedings of the 13th IEEE/ACM International Symposium on Modeling, Analysis, and Simulation of Computer Telecommunication Systems (MASCOTS), pages 135-142, Atlanta, September 2005.

  [26]

  Q. Zhang, A. Riska, E. Riedel, ``Workload propagation - Overload in bursty servers'', in Proceedings of the 2nd International Conference on Quantitative Evaluation of Systems (QEST), pages 179-188, Torino, Italy, September 2005.

  [27]

  J. Zhang, A. Sivasubramaniam, A. Riska, Q. Wang, E. Riedel, ``An Interposed 2-Level I/O Scheduling Framework for Performance Virtualization'', in Proceedings of ACM SIGMETRICS, extended abstract, pages 406-407, Banff, Canada, June 2005.

  [28]

  Q. Zhang, A. Riska, E. Smirni, E. Riedel, ``Bottleneck Identification in E-commerce Systems'', in Proceedings of the 9th International Workshop on Web Content Caching and Distribution (WCW), pages 273-282, Beijing, China, October 2004.

  [29]

  A. Riska, E. Riedel, S. Iren, ``Adaptive disk scheduling for overload management'', in Proceedings of the 1st International Conference on Quantitative Evaluation of Systems (QEST), pages 176-186, Enshede, The Netherlands, September, 2004.

  [30]

  A. Riska, E. Riedel, ``It's Not Fair - Evaluating Efficient Disk Scheduling'', in Proceedings of the 11th IEEE/ACM International Symposium on Modeling, Analysis, and Simulation of Computer Telecommunication Systems (MASCOTS), pages 288-295, Orlando, FL, October 2003.

  [31]

  A. Riska, E. Riedel, S. Iren, ``Managing Overload Via Adaptive Scheduling'', in Proceedings of the 1st Workshop on Algorithms and Architecture for Self-Managing Systems, pages 23-24, San-Diego, CA, June 2003.

  [32]

  A. Riska, E. Smirni, and G. Ciardo, ``Exact analysis of a class of GI/G/1-type performability models'', ACM SIGMETRICS Performance Evaluation Review, Vol. 31, Issue 2, pages 28-30, Dec 2003. Also presented at the Fifth Workshop on Mathematical Performance Modeling and Analysis (MAMA), San Diego, CA, June 2003.

  [33]

  A. Riska, V. Diev, E. Smirni, ``Efficient fitting of long-tailed data sets into hyperexponential distributions'', in Proceedings of IEEE Internet Performance Symposium (GlobeCom), IEEE Catalog Number 02CH37398C, Taipei, Taiwan, November 2002.

  [34]

  A. Riska, W. Sun, E. Smirni, G. Ciardo, ``ADAPTLOAD: effective balancing in clustered Web servers under transient load conditions'', in Proceedings of the 22nd International Conference on Distributed Computing Systems (ICDCS), pages 104-112, Vienna, Austria, July 2002.

  [35]

  A. Riska, M. Squillante, S.Z. Yu, Z. Liu, L. Zhang, ``Matrix-analytic analysis of a MAP/PH/1 queue fitted to Web server data'', in Matrix-Analytic Methods; Theory and Applications, G. Latouche and P. Taylor (editors), pages 333-356, World Scientific, Adelaide, Australia, July 2002.

  [36]

  A. Riska, E. Smirni, ``Exact aggregate solutions for M/G/1-type Markov processes'', in Proceedings of ACM SIGMETRICS, pages 86-96, Marina Del Rey, CA, June 2002.

  [37]

  A. Riska, V. Diev, E. Smirni, ``Efficient fitting of long-tailed data sets into phase-type distributions'', ACM SIGMETRICS Performance Evaluation Review, Vol.30, No. 3, pages 6-8. Also presented at the Fourth Workshop in Mathematical Performance Modeling and Analysis (MAMA), Marina Del Rey, CA, June 2002.

  [38]

  A. Riska, E. Smirni, ``MAMSOLVER: A matrix-analytic methods tool'', in T. Field et al. (editors), TOOLS: Lecture Notes in Computer Science 2324, pages 205-211, Springer-Verlag, London, April 2002.

  [39]

  D. Chen, A. Messer, P. Bernadat, G. Fu, Z. Dimitrijevic, D. Lie, D. Mannaru, A. Riska, D. Milojicic, ``JVM Susceptibility to Memory Errors'', in Proceedings of the USENIX Java Virtual Machine Research and Technology Symposium, pages 67-78, Monterey, CA, April 2001.

  [40]

  A. Riska, E. Smirni, G. Ciardo ``Analytic modeling of load balancing policies for tasks with heavy-tailed distributions'', in Proceedings of the 2nd ACM International Workshop on Software and Performance (WOSP), pages 147-157, Ottawa, Canada, September 2000.

  [41]

  G. Ciardo, A. Riska, E. Smirni, ``An aggregation-based solution method for M/G/1-type processes'', in Proceedings of the 3rd International Conference on Numerical Solutions of Markov Chains (NSMC), pages 21-40, Zaragoza, Spain, September 1999.

  OTHER REFEREED AND INVITED PUBLICATIONS

  [42]

  E. Smirni, Q. Zhang, N. Mi, A. Riska, and G. Casale, ``New Results on the Performance Effects of Autocorrelated Flows in Systems'', in the Workshop on Next Generation Software, co-located with IPDPS, Long Beach, CA, March 2007.

  [43]

  A. Riska, E. Riedel, Guest editing ``Bigger and Faster and Smaller'' for ACM SIGMETRICS Performance Evaluation Review, Vol. 33, No. 4, Guest Editor's foreword pages 2-3, March 2006.

  [44]

  A. Riska, E. Smirni, ``M/G/1-type Markov processes: A tutorial'', in M. C. Calzarossa and S. Tucci (editors), ``Performance Evaluation of Complex Computer Systems: Techniques and Tools'', Lecture Notes in Computer Science 2459, pages 36-63, Springer-Verlag 2002.

  PATENTS

  • [P1] A. Riska, N. Mi, E. Riedel, E. Smirni, ``Data Storage Device with Histogram of Idle Time and Scheduling of Background and Foreground Jobs'', patent pending, submitted November 2007.
  • [P2] A. Riska, E. Riedel, "Admission control at the disk drive'', patent pending, submitted June 2005.
  • [P3] S. Iren, A. Riska, E. Riedel, ``A Semantically-aware, dynamic, window-based disk scheduling method and apparatus for multimedia applications'', US patent 7,073,021 B2, granted on July 4, 2006, submitted March 2003.

  CONFERENCE TUTORIALS

  • 2005: "Performance Evaluation Issues in Storage System'', QEST'05, with Eitan Bachmat.
  • 2002: "Matrix-analytic methods as tools for solving M/G/1-type Markov processes'', ACM SIGMETRICS'02, with Evgenia Smirni.
  • 2002: ``M/G/1-type Markov processes'', IFIP Group 7.3 Conference on Performance, PERFORMANCE'02, with Evgenia Smirni.

  SOFTWARE

  • 2002: MAMSolver: A software tool for the solution of different types of analytic models used for the performance evaluation of computer and communication systems (tool web-site http://www.cs.wm.edu/MAMSolver/).

  PROFESSIONAL ACTIVITIES: REVIEWING

  • NSF Panel 2006.
  • Journals

    IEEE Transactions on Computers, Parallel and Distributed Systems, Reliability, Software Engineering, Dependable and Secure ComputingDependable and Secure Computing; Journal of ACM; ACM Transactions on the Web; Elsevier Performance Evaluation Journal; Journal of Systems and Software; Journal of Operations Research; Annals of Operations Research; Journal of Queueing Systems (QUESTA); International Journal of Communication Systems;

  • Conferences

    USENIX Conference on File and Storage Technologies (FAST) 2003, 2005, 2007; International Symposium on Performance Analysis of Systems and Software (ISPASS) 2006; Symposium on Dependable Systems and Networks (DSN) 2005, 2006; ACM SIGMETRICS Conference 2001, 2003, 2005, 2007; IFIP 7.3 International Symposium on Computer Performance, Modeling, Measurements and Evaluation (PERFORMANCE) 2002, 2005; Supercomputing 2005; ACM SIGMETRICS Workshop in Software and Performance (WOSP) 2002; International Parallel and Distributed Processing Symposium (IPDPS) 2002; Modeling Techniques and Tools for Computer Performance Evaluation (TOOLS) 2002.

  OTHER PROFESSIONAL ACTIVITIES

  • EYH: Expanding Your Horizons in Science and Mathematics Conference: for nurturing young girls' interest in science and math courses. Participated as a workshop leader in the regional EYH conference organized at the Robert Morris University in 2007 and 2008.

  AWARDS

  • 2002: Stephen K. Park Research Award, Computer Science Department, College of William and Mary, Williamsburg, VA.
  • 1995: EU TEMPUS Scholarship, for excellent achievements during undergraduate studies, University of Tirana, Albania.
  • 1989: Winner of the National Olympiad on Mathematics for high school students, Tirana, Albania.