Katherine Barabash
Home: +972-4-8324821
IBM Office: +972-4-8296573
Technion Office: +972-4-8294619
kathy@il.ibm.com
|
Qualifications |
Systems design and programming. Computer related research. Small-scale software projects management. · Programming languages: Java, C, modeling and scripting languages · Operating systems: Windows, Unix (AIX), Linux · Development environments: Unix, Eclipse, Rational, Microsoft (Visual, .NET) |
|
Experience |
1997 – Current IBM Research 2008 – 2009. Relocation Enablers for Power – Project Leader In this position I was responsible for the research effort
to allow boundary less live migration of logical partitions in IBM PowerVM systems. We have investigated the existing
technology restrictions and have decided to concentrate first on allowing
live partition migration across network subnet boundaries. The solution was
specified, designed and implemented by a team of two researches and a student
and has become part of the demonstration presented at CeBIT
2009 technology fair in The project has a continuation into 2009, with a focus on migrating partition’s storage thus removing a requirement for the shared storage disks for participating systems. I was responsible for formulating the 2009 research proposal and goals but will not be lading the project in 2009 because of my M.Sc. studies this year.
2004 – 2008. Storage Server Integration - Project Leader In this position I was responsible for specification, design, prototyping and development of Server Storage Provisioning Tool (SSPT). SSPT is a Java library to provision SAN storage to servers built to automate complex SAN configuration tasks in heterogeneous multi vendor environment. Working on this project has build a strong knowledge base and experience in the area of storage configuration and management, familiarity with different vendors’ SAN related products and involvement with emerging related standards (CIM, SMIS, iSCSI), protocols (SLP, CIM over HTTP) and technologies (FC and iSCSI SAN boot, blade server failover). SSPT core team consisted of two to four research staff members over the years. My responsibility was coordinating the work of this core team and maintaining relationships with larger development teams owning IBM products where SSPT is embedded (RDM , IBM Director, IBM Systems Director). |
|
|
2002 – 2004. Dynamic and Adaptive JVM - Project leader In this research projects I was coordinating the work of two person team aimed at making the JVM runtime resource utilization manageable. The work was based on a production IBM JVM. We have designed, build and benchmarked (on several platforms) a prototype of adaptive JVM, one that is aware of other JVMs running in the system and adapting its resource utilization (mainly in GC area – amount of threads spawned by the GC, timing of minor and major GC pauses, timing of heap compactions). This effort was an independent part of larger IBM research project called Work Load Manager (WLM). |
|
|
2002 – 2004. Concurrent GC performance tuning project – research team member This short-term research project was aimed at identifying and eliminating one of the performance hits of the concurrent Garbage Collection algorithms. I was responsible for developing and following the measurements methodology to identify the source of the problem and in implementing and benchmarking the various proposed solution variants. Along with other team members I was involved in building the final solution and making it part of the IBM JVM product. |
|
|
1999 - 2001. Cache Conscious Memory Management project – research team member This research project was aimed at building a prototype implementing an idea from an influential memory management research paper in a production level IBM JVM. The idea was to make the memory allocation and reclaiming conscious of data placement in a processor cache. I was responsible for complementary research, benchmarking and implementation. This project ended up with us being unable to match the result reported by a research paper in a highly optimized commercial JVM. The project was followed by filing the technical report. Working on this project I was involved with MMU performance measurements for both Power and Intel architectures. |
|
|
1997 – 1999. Concurrent GC project – graduate student position In this position I was working on a team developing the concurrent Garbage Collection for IBM JVM. During this time, I was given the following responsibilities:
|
|
Education |
2007 – current Technion - 1994 – 1997 Technion - 1991 – 1994 Technion - BS in Applied Mathematics, Cum Laude 1988 - 1990 1977 - 1988 School education |
|
Awards |
2000. Research Division Award for Contribution to JDK 1.3.0 for S/390 2003. First Patent Application Invention Achievement Award for Mostly Concurrent Garbage Collection 2004. Research Division Award for Contribution to the Garbage Collection for large heaps |
|
Publications |
·
Tamar Domani, Elliot K. Kolodner,
Ethan Lewis, Eliot E. Salant,
Katherine Barabash, Itai Lahan, Yossi Levanoni, Erez Petrank, Igor Yanover:
Implementing an On-the-Fly Garbage Collector for Java. ISMM 2000:
155-166 · Katherine Barabash, Yoav Ossia, Erez Petrank: Mostly concurrent garbage collection revisited. OOPSLA 2003: 255-268 · Katherine Barabash, Niv Buchbinder, Tamar Domani, Elliot K. Kolodner, Yoav Ossia, Shlomit S. Pinter, Janice C. Shepherd, Ron Sivan, Victor Umansky: Mostly Accurate Stack Scanning. Java™ Virtual Machine Research and Technology Symposium 2001 · Katherine Barabash, Yoav Ossia, and Erez Petrank. Mostly Concurrent Garbage Collection Revisited. Proceedings of the ACM Conference on Object-Oriented Programming, Systems, Languages, and Applications (OOPSLA'03), October 2003. · Katherine Barabash, Ori Ben-Yitzhak, Irit Goft, Elliot K. Kolodner, Victor Leikehman, Yoav Ossia, Avi Owshanko, Erez Petrank: A parallel, incremental, mostly concurrent garbage collector for servers. ACM Trans. Program. Lang. Syst. 27(6): 1097-1146 (2005) |
|
Additional Information |
Born in Spoken languages: Russian, English, Hebrew |