System Engineer
Resume:
Lei Xu Resume
Lei Xu
*th Year Computer Science Ph.D Candidate
Computer Science and Engineering
University of Nebraska-Lincoln
Email: ******@*****.***
Webpage: http://cse.unl.edu/ lxu/
Github: https://github.com/eddyxu
Mobile: 402-***-****
Research Interests
Large Scale Search-able File System, File System Performance on Manycore and SCM systems, Dis-
tributed Storage Systems, Cloud Computing.
Education
Ph.D Student, Computer Science (GPA 3.91) 09/2008 09/2013
University of Nebraska-Lincoln
Advisor: Hong Jiang and David Swanson
09/2001 06/2005
Bachelor of Engineering, Electronic Science and Technology
Wuhan University, Hubei, China.
09/2001 06/2005
Bachelor of Engineering, Computer Science and Technology
Wuhan University, Hubei, China.
Academic and Professional Experiences:
Software Engineering Intern at Google Summer 2012
Worked in one backend storage team for Google Ads. I have worked on several optimization projects,
including improving the CPU utilization of the one large-scale Ads backend system, and automatically
examining the correctness of several data processing algorithms used by Google Ads.
1
Lei Xu Resume
Software Engineering Intern at Google Summer 2011
Designed and implemented a log-structured back-end storage for a globally distributed cloud product.
The log-structured storage system was designed to save the bandwidth of propagating data between
data centers.
Fall 2009 present
Research Assistant
Professor Hong Jiang, University of Nebraska-Lincoln
Topics:
Large Scale Search-able File System. To address the challenges of fast retrieving data from large
le systems, we design a distributed le system architecture that supports performing real-time
inline le-index and le-search functions. To overcome the scalability issue of the high indexing
overhead, a novel access-based partitioning mechanism is proposed to leverage the access locality
to improve the le index/search performance.
Virtual File System(VFS) Scalability on Manycore and Storage Class System (SCM). We eval-
uate the Linux VFS performance on a 48-core AMD machine with RAM-speed storage devices
and nd that the in-RAM computation of VFS will become the I/O bottleneck. To address this
performance issue, I design a distributed directory cache (dcache) in Linux kernel, as an example
to improve the VFS scalability, that explore the access locality of each CPU core in order to
reduce the cache coherency overhead and multi-core spin-lock overhead.
Fall 2008 Summer 2009
Research Assistant
Professor David Swanson, University of Nebraska-Lincoln
Topic: Performance and Reliability Analyzing for Cluster Storage of High-End Computing.
I worked on customizing Hadoop Distributed File System (HDFS) to support the requirements
of the Holland Computing Center (HCC) at UNL. Several patches were accepted by the upstream
Hadoop community.
March 2007 May 2008
Co-founder and Software Architect Quarkware.com
I had designed and implemented a distributed le system, inspired from Google File System, by using
C++ and ACE.
2
Lei Xu Resume
February 2006 February 2007
Software Engineer Roxbeam.com
I worked in the P2P VOD team and wrote several P2P backend-servers (e.g., le chunk serving, billing
and etc.) in the P2P VOD system, which was used by Yahoo Japan to streaming several sport events.
I also wrote a cluster monitor system to measure the healthy of this VOD system.
July 2005 January 2006
Storage Optimizing Engineer Sina.com
Designed and implemented a distributed storage migrating system, which monitors the healthy status
and utilizations of the entire e-mail system and provided strategies of balancing the storage between
servers.
Selected Publications
1. Lei Xu, Hong Jiang, Xue Liu, Lei Tian, Yu Hua, and Jian Hu, Propeller: A Scalable Metadata
Organization for A Versatile Searchable File System, CSE Technique Report.
2. Lei Xu, Jian Hu, Stephen Mkandawire, Hong Jiang, SHHC: A Scalable Hybrid Hash Cluster
for Cloud Backup Services in Data Centers., DCPerf 2011.
3. Jian Hu, Hong Jiang, Lei Tian, Lei Xu, PUD-LRU: An Erase-E cient Write Bu er Manage-
ment Algorithm for Flash Memory SSD, MASCOTS 2010.
4. Lei Xu, Hong Jiang, Lei Tian, Hi re: A High Fidelity Trace Replayer for Large-Scale Storage
Systems, Work-in-Progress in FAST 10.
Major abilities:
Programming Languages
Expert in modern C++ design.
Expert in Python (10 years).
Others
Strong knowledges of distributed storage and le system design principles.
Strong knowledges about Google File System/Hadoop Distributed File System (HDFS),
Ext3/4 le system. I have contributed several accepted patches to HDFS.
Strong knowledges of Linux Virtual File System, FUSE and Operating Systems.
Strong knowledges of multicore/manycore architectures (NUMA, cache hierarchy and etc.)
3
Lei Xu Resume
Familiar with MapReduce, Bigtable, MegaStore, distributed computing, Chubby, Stubby,
F1.
10+ Years experiences of Linux development.
Experienced with Linux Kernel, Ext2/3, Btrfs, ZFS.
Experienced with RAID, Solid State Disk(SSD), SCM, De-duplication technologies.
Strong knowledges about Network Programming, Multi-Threads/Multi-Core/Many-Cores
Programming.
Good at algorithms, data structures, and design patterns.
Personal Activities
Membership in Professional Societies
Student Member, USENIX
Awards
USENIX Student Grent, 8th USENIX Conference on File and Storage Technologies(FAST 10)
4
Contact this candidate