Piyush Shivam   Computer Science Department Duke University Email: shivam AT cs dot duke dot edu Office phone: 919-660-6586

Education

• PhD, Computer Science, Duke University, 2007
  Dissertation: Proactive Experiment-Driven Learning for System Management
  Advisors: Dr. Jeff Chase and Dr. Shivnath Babu
  
  
• MS, Computer and Information Science, The Ohio State University (OSU), 2002
  Thesis: High Performance User Level Protocol on Gigabit Ethernet
  Advisor: Dr. Dhabaleswar Panda
  
  
• MSc.(Tech), Information Systems, Birla Institute of Technology and Science(BITS), Pilani, INDIA, 1999
  

Research

I am interested in designing, building, and evaluating systems. My research spans across autonomic computing, network based computing, storage systems, statistical learning, performance modeling, high performance computing, Web services, and databases.

• Research Summary

  Addressing system management problems is one of the key challenges in computer science research. Examples include system tuning, performance debugging, availability management, anomaly detection, and capacity planning. A range of application-, hardware-, and software-related factors contribute to a potentially large factor-space that affects system behavior. A careful understanding of these factors is crucial for making correct system management decisions.
  
  My research involves design and development of policies and mechanisms to understand the impact of all the relevant factors that affect system behavior, and how these factors interact with one another. The goal is to create this knowledge in a proactive, automatic, efficient, and accurate manner. In my graduate research, I have used these policies and mechanisms to address system management problems in Storage Systems, High Performance Computing (HPC), and Web Services. Selected projects are summarized here.

• Publications

  
  
  • Proactive Experiment-Driven Learning for System Management. Piyush Shivam, Ph.D. Dissertation.
    
    
  • Cutting Corners: Workbench Automation for Server Benchmarking. Piyush Shivam, Varun Marupadi, Jeff Chase, Thilee Subramaniam, and Shivnath Babu. To appear in USENIX Annual Technical Conference. June, 2008.
    
    
  • Towards an Autonomic Computing Testbed. Aydan Yumerefendi, Piyush Shivam, David Irwin, Pradeep Gunda, Laura Grit, Azbayar Demberel, Jeff Chase, and Shivnath Babu. Workshop on Hot Topics in Autonomic Computing (HotAC II), held in conjunction with International Conference on Autonomic Computing (ICAC). June 2007.
    
    
  • Harnessing Virtual Machine Resource Control for Job Management. Laura Grit, David Irwin, Varun Marupadi, Piyush Shivam, Aydan Yumerefendi, Jeff Chase, and Jeannie Albrecht. Workshop on System-level Virtualization for High Performance Computing (HPCVirt), held in conjunction with EuroSys. March 2007.
    
    
  • Active and Accelerated Learning of Cost Models for Optimizing Scientific Applications. Piyush Shivam, Shivnath Babu, and Jeff Chase. International Conference on Very Large Data Bases (VLDB). September 2006.
    
    
  • Active Sampling for Accelerated Learning of Performance Models. Piyush Shivam, Shivnath Babu, and Jeff Chase. First Workshop on Tackling Computer Systems Problems with Machine Learning Techniques (SysML). June 2006.
    
    
  • Learning Application Models for Utility Resource Planning. Piyush Shivam, Shivnath Babu, and Jeff Chase. IEEE International Conference on Autonomic Computing (ICAC). June 2006.
    
    
  • ACE: Classification for Information Lifecycle Management. Gauri Shah, Kaladhar Voruganti, Piyush Shivam, and Maria Alvarez. IEEE Conference on Mass Storage Systems and Technologies (MSST). May 2006. (IBM Technical Report)
    
    
  • Model-Driven Placement of Compute Tasks and Data in a Networked Utility . Piyush Shivam, Adriana Iamnitchi, Aydan Yumerefendi, and Jeff Chase. IEEE International Conference on Autonomic Computing (ICAC). July 2005 (A short paper).
    
    
  • On the Elusive Benefits of Protocol Offload. Piyush Shivam and Jeff Chase. ACM SIGCOMM workshop on Network-I/O Convergence: Experience, Lessons, Implications (NICELI). August 2003.
    
    
  • High Performance User Level Sockets over Gigabit Ethernet. Pavan Balaji, Piyush Shivam, Pete Wyckoff, and Dhabaleswar K. Panda. IEEE International Conference on Cluster Computing (CC). September 2002.
    
    
  • Can User Level Protocols Take Advantage of Multi-CPU NIC?. Piyush Shivam, Pete Wyckoff, and Dhabaleswar K. Panda. IEEE/ACM International Parallel and Distributed Processing Symposium (IPDPS). April 2002.
    
    
  • EMP: Zero-copy OS-bypass NIC-driven Gigabit Ethernet Message Passing. Piyush Shivam, Pete Wyckoff, and Dhabaleswar K. Panda. ACM/IEEE Conference on Supercomputing (SC). November 2001.

• Demonstrations

  • Automated and On-Demand Provisioning of Virtual Machines for Database Applications . Piyush Shivam, Azbayar Demberel, Pradeep Gunda, David Irwin, Laura Grit, Aydan Yumerefendi, Shivnath Babu, and Jeff Chase. ACM International Conference on Management of Data (SIGMOD). June 2007.
    
    
  • ACE: Classification for Information Lifecycle Management. Gauri Shah, Kaladhar Voruganti, Piyush Shivam, and Maria Alvarez. IBM Almaden Research Center. June 2005.

• Presentations

  • Experiment-Driven Management of Web Services.
    International Conference on the Virtual Computing Initiative (VCI). Raleigh, NC, May 2007. [ppt]
  • Model-Driven Placement of Compute Tasks and Data in a Networked Utility.
    IEEE International Conference on Autonomic Computing (ICAC). Seattle, Washington, June 2005. [ppt]
  • On the Elusive Benefits of Protocol Offload.
    ACM SIGCOMM workshop on Network-I/O Convergence: Experience, Lessons, Implications (NICELI). Karlsruhe, Germany, August 2003. [ppt]
  • Can User Level Protocols Take Advantage of Multi-CPU NICs?
    ACM/IEEE Parallel and Distributed Processing Symposium. Fort Lauderdale, Florida, April 2002. [ppt]
  • Low Latency, High Bandwidth Messaging Layer on Gigabit Ethernet.
    ACM/IEEE Conference on SuperComputing. Denver, Colorado, November 2001. [ppt]

• Patent

  • OffloadMagic: A Model and System for a Self-configuring and Adaptive Offload Card Architecture for TCP/IP and specialized protocols, 2003.

    The benefits of offloading network protocol processing to the NIC are highly application dependent. In fact, for some applications NIC offload can lead to degraded performance. This patent presents techniques that adapt the protocol offload to a given application. This work was done at IBM Almaden Research Center with Sandeep Uttamchandani and Kaladhar Voruganti.

• Relevant Coursework

  • Federated Distributed Systems
  • Machine Learning
  • Statistical Learning Theory
  • Performance, Reliability and Availability Modeling of Computer Systems
  • Advanced Computer Networks
  • Network Programming
  • Distributed Information Systems
  • Experimental Methods in Computer Systems
  • Advanced Computer Architecture
  • Design and Analysis of Algorithms
  • Introduction to High Performance Computing

Work Experience/Internships

• Sun Microsystems, Network File Systems (NFS), Broomfield, CO USA. Mentor: Doug McCallum, Manager: Don Traub. (05/06 - 07/06)
• IBM Almaden Research Center, Autonomic Storage Management, San Jose, CA USA. Mentor: Gauri Shah, Manager: Sandeep Gopisetty. (05/05 - 08/05)
• IBM Almaden Research Center, Storage Subsystems, San Jose, CA USA. Mentor: Renu Tewari, Manager: Norm Pass. (05/03 - 08/03)
• Dell Computers, Information Technology, Austin, Texas, USA. Manager: James Roberson. (06/01 - 08/01)
• Wipro Technologies, Bangalore, INDIA and IBM Beaverton, Oregon, USA. Manager: H R Venkatesh. (08/99 - 07/00)
• IBM Global Services, Bangalore, India. Manager: Rajendra K. Bera. (01/99 - 06/99)

Personal

Annika Elizabeth Shivam was born on March 16th 2007! Here are a few pictures.