Ph.D.		November, 2017
Institution		University of Pittsburgh (Pitt) -- Pittsburgh, U.S.
School		School of Computing and Information
Major		Computer Science
Dissertation		Adaptive and power-aware fault tolerance for future extreme-scale computing
Courses taken		Fall 2015: Advanced Topics in Distributed Real-time Systems Spring 2014: Principles of Database Systems, Advanced Topics in Operating Systems Fall 2013: Distributed Operating System, Theory of Computation, Multimedia Software Engineering Spring 2013: Wide Area Networks, , Wireless Networks Fall 2012: Computer Architecture, Artificial Intelligence, Non-linear Optimization
GPA		3.88/4





B.Eng.		July, 2012
Institution		Xi'an Jiaotong University(XJTU) -- Xi'an, China
School		Electronic and Information Engineering
Major		Computer Science and Technology
Courses taken		Computer Networks, Artificial Intelligence, Software Engineering, Microcomputer Interface, Computer Architecture, Database Systems, Operating System, Compiler, Computer Composition, X86 Assembly Language, Data Structure and Algorithms, Discrete Mathematics, Probability and Mathematical Statistics, Digital Logic and System, Signals and Systems, Linear Algebra and Analytic Geometry of Space, Advanced Mathematics
GPA		90.14/100(overall) 92.44/100(major)
Rank		2/145

Research interests

• Parallel computing
• Large scale distributed systems, focusing on energy efficiency and fault tolerance
• High Performance Computing and Cloud Computing

Dissertation

Publications

• Xiaolong Cui, Taieb Znati, and Rami Melhem. "Rejuvenating Shadows: Fault tolerance with Forward Recovery." The 19th IEEE International Conference on High Performance Computing and Communications (HPCC 2017), Bangkok, Thailand, 2017.[pdf]
• Xiaolong Cui, Taieb Znati, and Rami Melhem. "Adaptive and Power-Aware Resilience for Extreme-scale Computing." in Frontiers in Signal Processing, Volume 1, Number 1, July 2017.
• Xiaolong Cui, Taieb Znati, and Rami Melhem. "Adaptive and Power-Aware Resilience for Extreme-scale Computing." The 16th IEEE International Conference on Scalable Computing and Communications (ScalCom2016). July 18-21, 2016 (Best Paper Award) [pdf]
• Xiaolong Cui, Tariq Alturkestani, Esteban Meneses, Taieb Znati and Rami Melhem. Leaping Shadows: Adaptive and Power-aware Resilience for Extreme-scale Systems. The European Conference on Computer Systems (EuroSys). London, UK, 2016 (Poster) [pdf]
• Xiaolong Cui, Bryan Mills, Taieb Znati, and Rami Melhem. "Shadow Replication: An Energy-Aware, Fault-Tolerant Computational Model for Green Cloud Computing." Energies 7, no. 8 (2014): 5151-5176. [pdf]
• Xiaolong Cui, Bryan Mills, Taieb Znati, and Rami Melhem. "Shadows on the Cloud: An energy-aware, profit maximizing resilience framework for cloud computing." International Conference on Cloud Computing and Services Science(CLOSER). April 3-5, 2014. [pdf]

Scalable, power-aware fault tolerance approaches for extreme-scale systems (May 2013 -- Nov. 2017)

• Project description: the U.S. Department of Energy has identified resilience and power consumption as key challenges for future extreme-scale systems. It is commonly thought that current checkpoint/restart methods don't scale. This project aims at proposing new fault tolerance scheme to minimize power as well as increase scalability for future extreme-scale system.
  
• Accomplishments:
      - Proposed a new power-aware fault tolerance approach, referred to as Leaping Shadows, as part of Shadow Computing, for future extreme-scale systems
      - Developed a profit based framework to predict the profitability of above approach
      - Implemented a prototype of Leaping Shadows in MPI
      - Evaluated the performance of above approach in both HPC and Cloud environments

  Advisor : Taieb Znati, Rami Melhem

  
  

Analytics database system with transactional support (March 2014)

• Project description: the goal of this project is to provide transactional support for a small analytics database system. Specifically, the objective is to develop PittHN that efficiently support OLTP (i.e., transactions) and OLAP (i.e., aggregate queries) workloads. Basically, PittHN implements SAP HANA’s L1-delta and L2-Delta.
  
• Accomplishments:
      - Designed a database system that supports concurrent access to data by both transactions and queries while ensuring serializability
      - Implemented a Transaction Manager that reads commands from different program files concurrently and pass the commands to the Scheduler
      - Implemented a Scheduler with a lock manager for Strict Two-Phase Locking protocol and a deadlock detector
      - Implemented a Data Manager for maintaining data in tables with both row store and column store
  

MiniGoogle for documents indexing and search (Nov. 2013 -- Dec. 2013)

• Project description: Increasingly, large-scale Internet services are being deployed atop multiple geographically distributed data centers. These services must scale across a large number of machines, tolerate failures, and support a large volume of concurrent requests. The main objective of this project is to design and implement a basic data-intensive application to index and search large documents.
  
• Accomplishments:
      - Used socket programming to implement a client-server system with DNS server
      - Designed and implemented a distributed and multithreaded algorithm to index and search large documents
      - Implemented the above system with Hadoop MapReduce framework as an alternative approach

Light weight user-level thread (LWT) scheduling system (Sep. 2013 -- Oct. 2013)

• Project description: The purpose of this project is to implement a light weight user-level thread (LWT) scheduling system without the need to modify the Linux kernel. Your system must allow the creation and concurrent execution of threads of control within a Linux task, in a true preemptive manner. The LWT system must be capable of creating threads and performing mini context switches to share the CPU time among the task threads, based on the specified scheduling policy.
  
• Accomplishments:
      - Designed and implemented a library for thread creation, wait, sleep, and exit
      - Implemented priority-based thread scheduling using SIGALRM handler and implement semaphores for mutex
      - Demonstrated the correctness of above system with producer-consumer problem

Simplified File Transfer Protocol (FTP) (Mar. 2013 -- Apr. 2013)

• Project description: The purpose of a file transfer protocol is to enable the transfer of files between machines, typically under the command of a user. Several issues need to be addressed in the design of a file transfer protocol, including dealing with differences in file name conventions, text and data representation, and directory structure. Furthermore, the protocol must ensure reliable transfer of files from one system to another. The main objective of this project is to design and implement a Simple File Transfer Protocol (SFTP) on top of an unreliable transport channel.
  
• Accomplishments:
      - Designed and implemented a simplified FTP protocol with both client and server using layered architecture
      - Implemented and evaluate multiple ARQs by injecting errors and packet drops
      - Supported concurrent access and file transfer

Innovative page replacement algorithm for database on Non-Uniform Storage Architecture (NUSA) (Feb. 2013 -- April 2013)

• Project description: It is common that enterprise business servers use hybrid storage devices (NUSA) in their system. For example, a system can have both HDD and SSD installed. In such situation, however, the existing page replacement algorithms may not work well for database system in terms of overall system performance, fairness, scalability, and so on. This project explores new page replacement algorithm for database on NUSA.
  
• Accomplishments:
      - Implemented an innovative page replacement algorithm called Scanning Group (SG) in MySQL
      - Tested the performance and overhead of SG on a system with both SSD and HDD

Simulator for distributed directory cache coherence protocols in CMPs (Nov. 2012)

• Project description: In this project, you will write a simulator to evaluate MSI and MESI snooping cache coherence protocols in CMPs with multilevel caches and 32-bit word size. You will assume write-back (rather than write through) with invalidate (rather than update) protocols.
  
• Accomplishments:
      - Designed and implemented MSI cache coherence protocol for CMP with private L1 cache and shared distributed L2 cache
      - Simulated the delay of a mess NoC
      - Performed comparative analysis among different configures for multiple benchmarks

Simulation of dynamically scheduled processor with MIPS64 ISA (Oct. 2012)

• Project description: The purpose of this programming assignment is to evaluate this effect by simulating a modified (and simplified) version of the PowerPc 604 and 620 architectures. Your simulator should take an input file as a command line input, recognize the different fields of the instructions and simulate their execution.
  
• Accomplishments:
      - Designed and implemented Tomasolo algorithm with renaming registers and re-order buffer
      - Implemented branch target buffer for dynamic branch prediction
      - Evaluated the performance of different configurations for multiple benchmarks

IEEE ScalCom best paper award		2016		Research competition winner		2016
Research competition winner		2015		People's Choice Award@Randall Family Big Idea Competition		2015
Dietrich School of Arts and Science Fellowship		2012		Excellent student leader		2011
National Lizhi Scholarship		2011		Excellent student leader		2010
PengKang Scholarship		2010		Excellent student		2009
PengKang Scholarship		2009		Excellent student leader		2007
First Prize in National Mathematics Olympiad (China)		2007		First Prize in National English Olympiad (China)		2004

Part-time Instructor		Fall 2016

Teaching CS1645/2045 (Intro to High Performance Computing) for undergraduate and graduate

VP, Graduate Student Organization		May 2016 -- April 2017

Organizing annual welcome picnic

Hosting weekly coffee hour

Volunteer		May 2015 -- present

Light of Life Rescue Mission

Devopsdays Pittsburgh 2015

Paper Reviewer		May 2013 -- present

ACM: Transactions on Embedded Computing Systems (TECS)

IEEE: Transactions on Computers (TC)

IEEE: International Green Computing Conference (IGCC)

Elsevier: Journal of Computer and System Sciences (JCSS)

Teaching Assistant		Jan. 2013 -- present

Computer architecture	(Fall 2013)
Intermediate programming using Java	(Spring 2012, Fall 2013)
Introduction to computer programming with Python	(Spring 2013)

Class President		Sep. 2009 -- July 2011

Won the Outstanding Class Prize with my classmates.

Hosted discussions in department and organized many class activities.

Accompanied a classmate who had a surgery through a night in hospital.