Brief Bio 
  • Youngjae Kim is a research staff member at Oak Ridge National Laboratory managed for the U.S Department of Energy by UT-Battelle, LLC. He has a broad research background with an in-depth knowledge in various computer systems area. Specifically his research interests are in parallel computing, storage systems, distributed systems, networked systems, non-volatile memory devices (SSDs). Among these areas, His current research is focused on parallel file sytsems, networking, and storage system. 

  • He is currently affiliated with Technology Integration Group at NCCS, Oak Ridge National Laboratory. NCCS hosts the Oak Ridge Leadership Computing Facility which is home to the Titan supercomputer. TechInt is the core R&D computer systems and software group in charge of delivering new solution in a variety of areas such as file and storage systems, data management, non-volatile memory, architecture, and networking for leadership-class HPC systems and works on several DOE funded R&D projects at ORNL. 
We have several internship opportunities available throughout the year or short-term, or summer
If you have innovative ideas on enhancing current HPC systems, please freely contact me at kimy1 {at} ornl {dot} gov.  

Research Area
  • Parallel and Distributed Systems, High Performance Computing, Operating Systems, HPC in Cloud, Computer Architecture, Visualization, Cloud, Big Data 
  • Specific fields of interest include
    • File Systems, Storage systems, Parallel I/O, Non-volatile Memory
    • Operating Systems, Distributed Systems, Networking
    • Data Management, Mathematical Modeling and Optimization 
    • Big Data - Data infrastructure (Storage and Networking) and Data Analytic Systems

Current Projects
  • Burst buffer + Checkpoint/Restart 
    • Explore various ways of the usage cases using SSDs for exa-scale systems (Checkpoint/Restart, Private Scratch/Swap, Global File System Cache, Data Sharing and In-transit analysis) 

  • Terabits networking project 
    • Developing an end-to-end data transfer application for bulk data movement  
      • Design data transfer protocol using CCI: multi-threaded implemented in pthreads in C, semaphores and locks 
        • Used more than tens of message types to implement efficient communication protocols between source and sink threads.
        • Used a shared RMA buffer for multiple worker threads to use it on the end system
    • Investigating an end-to-end data transfer optimization with layout-awareness on end-system hosts to match impedance between storage and network on the hosts.

  • Analysis-aware file system (2013-present) 
    • Developing ActiveFS, a new analysis-aware file system using active flash elements that are capable of computation. The active flash element can serve as a core component that can run data analysis tasks in storage in the Active File System. Active flash device will take advantage of idle cycles on its internal computing cores and high internal I/O bandwidth to run data analysis tasks therein. ActiveFS will rely on the workflow scheduling orchestration layer to minimize unnecessary data movements between the elements. 

  • NVM related projects 
    • Understanding the emerging storage technologies, helping architect next-generation compute and data intensive computing systems using the non-volatile memory devices. 
    • Specifically exploring non-volatile memory devices from several perspective; memory extension, fast storage buffer and out-of core data analysis.
      • NVMalloc project for memory extensions on NVM devices
      • DeepMap project for efficient variable-level placement on deep memory hierarchy
      • Active file system project for building data analysis services on storage 
    • Evaluating emerging storage and file system solutions such as hierarchical storage systems using hybrid disk and hybrid file systems. 

Past Research Projects
Education
Work History
Selected Publications (a complete list) (* corresponding author)
  • J. Lee, Y. Kim*, et. al., Synchronous I/O Scheduling of Independent Write Caches for an Array of SSDs, (to appear) IEEE Computer Architecture Letters (IEEE CAL)

  • Y. Kim, et. al., Layout-aware I/O Scheduling for Terabit Data Movement, IEEE Workshop on Distributed Storage Systems and Coding for Big Data, in conjunction with IEEE Big Data 2013, San Jose, CA, Oct 6-9, 2013 

  • Y. Kim, et. al., HybridPlan: A Capacity Planning Technique for Projecting Storage Requirements in Hybrid Storage SystemsSpringer Journal of Supercomputing (Springer J. Supercomputing), Vol. 67, No. 1, pp. 277-303, January 2014.

  • Y. Kim, et. al., Coordinating Garbage Collection for Arrays of Solid-state Drives, (In print) IEEE Transactions on Computers (IEEE TC), pp. 1-14, 2013. (DOI: 10.1109/TC.2012.256)

  • Y. Kim, et. al., A Temporal Locality-aware Page-Mapped Flash Translation Layer, Springer Journal of Computer Science and Technology  (Springer JCST), Vol. 28, No. 6, pp. 1025-1044, November 2013. 

  • J. Lee, Y. Kim*, et. al.Preemptible I/O Scheduling of Garbage Collection for Solid-state Drives, IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems (IEEE TCAD), Vol. 32, No. 2, pp. 247-260, February 2013. 

  • D. Tiwari, S. Boboila, S. Vazhkudai, Y. Kim, et. al., Active Flash: Towards Energy-Efficient, In-Situ Data Analytics on Extreme-Scale Machines Proceedings of the 11th USENIX Conference on File and Storage Technologies (FAST 2013), February, 2013. (24/127 = 18.9%)

  • S. Lim, J. Huh, Y. Kim, et. al., D-Factor: A Quantitative Model of Application Slow-Down in Shared Service Systems with Multiple Resources,  Proceedings of the ACM Int'l Conference on Measurement and Modeling of Computer Systems (SIGMETRICS 2012), London, United Kingdom, June 11-15, 2012. (31/203 = 15.2%)

  • Y. Kim, et. al., HybridStore: A Cost-Efficient, High-Performance Storage System Combining SSDs and HDDs, Proceedings of the 19th IEEE Int'l Symposium on Modeling, Analysis and Simulation of Computer and Telecommunication Systems (MASCOTS 2011), Singapore, July 25-27, 2011. (41/157 = 26.1%) 

  • R. Prabhakar, S. Vazhkudai, Y. Kim, et. al., Provisioning a Multi-Tiered Data Staging Area for Extreme-Scale Machines, Proceedings of the 31th Int'l Conference on Distributed Computing Systems (ICDCS 2011), Minneapolis, Minnesota, June 20-24, 2011. (87/565 = 15.1%) 

  • Y. Kim, et. al., Workload Characterization of a Leadership Class Storage, Proceedings of the 5th Petascale Data Storage Workshop (PDSW 2010), New Orleans, LA, November 2010. (Google Citations = 18 on Jan 15, 2014)

  • A. Gupta, Y. Kim, et. al., DFTL: A Flash Translation Layer Employing Demand-based Selective Caching of Page-level Address Mappings, Proceedings of the 14th Int'l Conference on Architectural Support for Programming Languages and Operating Systems (ASPLOS 2009), Washington DC, March 2009. (29/113 = 21.8%)

  • Y. Kim, et. al., Flashsim: A Simulator for NAND Flash-based Solid-state Drives, Proceedings of the Int'l Conference on Advances in System Simulation (SIMUL 2009), Porto, Portugal, September 2009. (Acceptance Rate: 31.0%) (Google Citations = 86 on Jan 15, 2014)  

  • Y. Kim, et. al., Managing Thermal Emergencies in Disk-Based Storage Systems, ASME Journal of Electronic Packaging (JEP), Volume 130, Issue 4, Pages 041105, 1-8, December 2008. (pdf) (Among top 10 most downloaded articles -- November 2008) 

  • J. Choi, Y. Kim, et. al., Modeling and Managing Thermal Profiles of Rack-mounted Servers with ThermoStat,Proceedings of the 13th Int'l Symposium on High Performance Computer Architecture (HPCA 2007), Phoenix, Arizona, February 2007. (28/174 = 16.0%) Best Paper Nominee (Google Citations = 36 on Jan 15, 2014)

  • Y. Kim, et. al., Understanding the Performance-Temperature Interactions in Disk I/O of Server Workloads, Proceedings of the 12th IEEE Int'l Symposium on High Performance Computer Architecture (HPCA 2006), Austin, Texas, February 2006. (26/172 = 15.6%) (Google Citations = 59 on Jan 15, 2014) 

Skills
  • Programming Languages: proficient in C, familiar with Python, Shell programming (bash), Matlab, Gnuplot, FORTRAN 90
  • Parallel Programming: pthread in C, MPI, distributed software shared memory (DSM) library 
  • Network programming: TCP/IP socket, Common Communication Interface (CCI)
  • Checkpointing Library: BLCR, Scalable Checkpoint Restart (SCR) library, OpenMPI checkpoint/restart library (ompi-checpoint, ompi-restart)
  • Parallel File Systems (PFS) and PFS utility: Lustre, LU (Lustre Utility) 
  • Benchmark programs
    • I/O benchmark: block-level performance (using linux aio library), Lustre performance benchmark, IOR, IOZONE, mpiBlast, FLASH IO
    • Scientific simulation programs: LAMMPS, S3D, GTC, GROMACS, QMCPACK, NAS benchmarks
    • CPU-memory benchmark: SPEC CPU2006
  • Memory profiling tools: valgrind (massif, memcheck)
  • Simulators
PhD Lineage