Where to go from here? Get real experience building systems! Opportunities: 496 projects –More projects:

Slides:



Advertisements
Similar presentations
Towards Automating the Configuration of a Distributed Storage System Lauro B. Costa Matei Ripeanu {lauroc, NetSysLab University of British.
Advertisements

Tapestry: Decentralized Routing and Location SPAM Summer 2001 Ben Y. Zhao CS Division, U. C. Berkeley.
1 StoreGPU Exploiting Graphics Processing Units to Accelerate Distributed Storage Systems NetSysLab The University of British Columbia Samer Al-Kiswany.
The Development of Mellanox - NVIDIA GPUDirect over InfiniBand A New Model for GPU to GPU Communications Gilad Shainer.
1 A GPU Accelerated Storage System NetSysLab The University of British Columbia Abdullah Gharaibeh with: Samer Al-Kiswany Sathish Gopalakrishnan Matei.
Priority Research Direction (I/O Models, Abstractions and Software) Key challenges What will you do to address the challenges? – Develop newer I/O models.
Development of a track trigger based on parallel architectures Felice Pantaleo PH-CMG-CO (University of Hamburg) Felice Pantaleo PH-CMG-CO (University.
Size Matters : Space/Time Tradeoffs to Improve GPGPU Application Performance Abdullah Gharaibeh Matei Ripeanu NetSysLab The University of British Columbia.
The Energy Case for Graph Processing on Hybrid Platforms Abdullah Gharaibeh, Lauro Beltrão Costa, Elizeu Santos-Neto and Matei Ripeanu NetSysLab The University.
1 The Case for Versatile Storage System NetSysLab The University of British Columbia Samer Al-Kiswany, Abdullah Gharaibeh, Matei Ripeanu.
1 Harvesting the Opportunity of GPU- based Acceleration Matei Ripeanu Networked Systems Laboratory (NetSysLab) University of British Columbia Joint work.
SAN DIEGO SUPERCOMPUTER CENTER at the UNIVERSITY OF CALIFORNIA; SAN DIEGO IEEE Symposium of Massive Storage Systems, May 3-5, 2010 Data-Intensive Solutions.
1 Harvesting the Opportunity of GPU-Based Acceleration for Data-Intensive Applications Matei Ripeanu Networked Systems Laboratory (NetSysLab) University.
1 Individual and Social Behavior in Tagging Systems Elizeu Santos-Neto David Condon, Nazareno Andrade Adriana Iamnitchi, Matei Ripeanu 20th ACM International.
Are P2P Data-Dissemination Techniques Viable in Today's Data- Intensive Scientific Collaborations? Samer Al-Kiswany – University of British Columbia joint.
1 stdchk : A Checkpoint Storage System for Desktop Grid Computing Matei Ripeanu – UBC Sudharshan S. Vazhkudai – ORNL Abdullah Gharaibeh – UBC The University.
WhereStore: Location-based Data Storage for Mobile Devices Interacting with the Cloud Patrick Stuedi, Iqbal Mohomed, Doug Terry Microsoft Research.
Beyond Music Sharing: An Evaluation of Peer-to-Peer Data Dissemination Techniques in Large Scientific Collaborations Thesis defense: Samer Al-Kiswany.
1 Exploring Data Reliability Tradeoffs in Replicated Storage Systems NetSysLab The University of British Columbia Abdullah Gharaibeh Matei Ripeanu.
Real Parallel Computers. Modular data centers Background Information Recent trends in the marketplace of high performance computing Strohmaier, Dongarra,
By Ravi Shankar Dubasi Sivani Kavuri A Popularity-Based Prediction Model for Web Prefetching.
1 Exploring Data Reliability Tradeoffs in Replicated Storage Systems NetSysLab The University of British Columbia Abdullah Gharaibeh Advisor: Professor.
A Workflow-Aware Storage System Emalayan Vairavanathan 1 Samer Al-Kiswany, Lauro Beltrão Costa, Zhao Zhang, Daniel S. Katz, Michael Wilde, Matei Ripeanu.
Computer System Architectures Computer System Software
Network Sharing Issues Lecture 15 Aditya Akella. Is this the biggest problem in cloud resource allocation? Why? Why not? How does the problem differ wrt.
IPlant Collaborative Tools and Services Workshop iPlant Collaborative Tools and Services Workshop Collaborating with iPlant.
ECE 526 – Network Processing Systems Design Network Processor Architecture and Scalability Chapter 13,14: D. E. Comer.
Chapter 2 Computer Clusters Lecture 2.3 GPU Clusters for Massive Paralelism.
11 If you were plowing a field, which would you rather use? Two oxen, or 1024 chickens? (Attributed to S. Cray) Abdullah Gharaibeh, Lauro Costa, Elizeu.
Min Xu1, Yunfeng Zhu2, Patrick P. C. Lee1, Yinlong Xu2
Emalayan Vairavanathan
1. 2 Corollary 3 System Overview Second Key Idea: Specialization Think GoogleFS.
Experience with Using a Performance Predictor During Development a Distributed Storage System Tale Lauro Beltrão Costa *, João Brunet +, Lile Hattori #,
1 Configurable Security for Scavenged Storage Systems NetSysLab The University of British Columbia Abdullah Gharaibeh with: Samer Al-Kiswany, Matei Ripeanu.
Ideas to Improve SharePoint Usage 4. What are these 4 Ideas? 1. 7 Steps to check SharePoint Health 2. Avoid common Deployment Mistakes 3. Analyze SharePoint.
Energy Prediction for I/O Intensive Workflow Applications 1 MASc Exam Hao Yang NetSysLab The Electrical and Computer Engineering Department The University.
Oracle Advanced Compression – Reduce Storage, Reduce Costs, Increase Performance Session: S Gregg Christman -- Senior Product Manager Vineet Marwah.
IPlant Collaborative Tools and Services Workshop iPlant Collaborative Tools and Services Workshop Collaborating with iPlant.
Directed Reading 2 Key issues for the future of Software and Hardware for large scale Parallel Computing and the approaches to address these. Submitted.
1 Multiprocessor and Real-Time Scheduling Chapter 10 Real-Time scheduling will be covered in SYSC3303.
SciDAC All Hands Meeting, March 2-3, 2005 Northwestern University PIs:Alok Choudhary, Wei-keng Liao Graduate Students:Avery Ching, Kenin Coloma, Jianwei.
Amy Apon, Pawel Wolinski, Dennis Reed Greg Amerson, Prathima Gorjala University of Arkansas Commercial Applications of High Performance Computing Massive.
A Measurement Based Memory Performance Evaluation of High Throughput Servers Garba Isa Yau Department of Computer Engineering King Fahd University of Petroleum.
1 Very similar items lost in the Web: An investigation of deduplication by Google Web Search and other search engines CWI, Amsterdam,
Microsoft Research1 Characterizing Alert and Browse Services for Mobile Clients Atul Adya, Victor Bahl, Lili Qiu Microsoft Research USENIX Annual Technical.
1 MosaStore -A Versatile Storage System Lauro Costa, Abdullah Gharaibeh, Samer Al-Kiswany, Matei Ripeanu, Emalayan Vairavanathan, (and many others from.
Parallelization and Characterization of Pattern Matching using GPUs Author: Giorgos Vasiliadis 、 Michalis Polychronakis 、 Sotiris Ioannidis Publisher:
Towards Exascale File I/O Yutaka Ishikawa University of Tokyo, Japan 2009/05/21.
Fast BVH Construction on GPUs (Eurographics 2009) Park, Soonchan KAIST (Korea Advanced Institute of Science and Technology)
Alok Choudhary Dept. of Electrical & Computer Engineering And Kellogg School of Management Northwestern University I/O and Storage: Challenges Moving Forward.
Next Generation Operating Systems Zeljko Susnjar, Cisco CTG June 2015.
Monte Carlo Data Production and Analysis at Bologna LHCb Bologna.
1 If you were plowing a field, which would you rather use? Two oxen, or 1024 chickens? (Attributed to S. Cray)
Jan 12, 2009LIGO-G Z1 DMT and NDS2 John Zweizig LIGO/Caltech Ligo PAC, Caltech, Jan 12, 2009.
DOE Network PI Meeting 2005 Runtime Data Management for Data-Intensive Scientific Applications Xiaosong Ma NC State University Joint Faculty: Oak Ridge.
Background Computer System Architectures Computer System Software.
COMP7500 Advanced Operating Systems I/O-Aware Load Balancing Techniques Dr. Xiao Qin Auburn University
CRISP WP18, High-speed data recording Krzysztof Wrona, European XFEL PSI, 18 March 2013.
Introduction to Performance Tuning Chia-heng Tu PAS Lab Summer Workshop 2009 June 30,
1
APE group Many-core platforms and HEP experiments computing XVII SuperB Workshop and Kick-off Meeting Elba, May 29-June 1,
Auburn University COMP8330/7330/7336 Advanced Parallel and Distributed Computing Parallel Hardware Dr. Xiao Qin Auburn.
University of Maryland College Park
Ricardo Jimenez-Peris Universidad Politecnica de Madrid
VDN: Virtual Machine Image Distribution Network for Cloud Data Centers
Degree-aware Hybrid Graph Traversal on FPGA-HMC Platform
A Software-Defined Storage for Workflow Applications
SDP Kernels Workshop – The Role of Kernels
Request Behavior Variations
Fast Accesses to Big Data in Memory and Storage Systems
Presentation transcript:

Where to go from here? Get real experience building systems! Opportunities: 496 projects –More projects: 491r/571r class next term –Theme: Harnessing massively multicore systems (e.g., GPUs) – USRA funding for summer –(applications due January 2011)

A few projects … The top 1% of searchers performs a full 13% of all searches in a given month. If you extend this to the top 20% the number of queries increase to roughly 70%." Read more at: " Although these relations are not news, the argument that follows and the link with advertisement revenues are really interesting. The goal of this project is to understand which queries are more likely to lead to advertising revenue using the search engine data Microsoft has (partially) made available [traces from their Live Search engine] Predict advertised item popularity?

Platform Example – Argonne Blue Gene/P 160K cores 10 Gb/s Switch Complex GPFS 24 servers IO rate : 8GBps= 51KBps / core !! 2.5K IO Nodes Torus Network 2.5 GBps per node 3D Torus 850 MBps per 64 nodes Tree The central storage is a bottleneck There are underutilized resources close to application

MosaStore Evaluation Overall: 1.52x DOCK6 Workflow Stages Read input, compute, and write temporary results Summarize, sort, and select Archive Versatile Storage Optimizations Cache the input data Cache temporary files Asynch. flush results to GPFS Results (8K processors) 1.06x 11.76x 1.51x Zhang et. al., “Design and Evaluation of a Collective I/O Model for Loosely- coupled Petascale Programming”, MTAGS ’09.

A few NetSysLab  P2P data storage system (MosaStore)  Application-level GPU harnessing  Online social systems

StoreGPU GPUs dramatically change the computation cost landscape.  10x FLOPS, 10x Memory bandwidth,  yet same cost! Q: Does the 10x reduction in computation costs GPUs offer change the way we design/implement (distributed) storage system? System design: balancing act in a multi-dimensional space.

Data deduplication System -- Prototype Evaluation Checkpointing a BLAST application 100 times 76% improvement in write throughput No negative impact on concurrent applications Throughput (MB/s) [S. Al-Kiswany, A. Gharaibeh, S. Gopalakrishnan, and M. Ripeanu, “A GPU Accelerated Storage System”, Submitted to NSDI ‘10.] no-SD SD-CPU SD-GPU

Characterizing Online Social Systems CiteULike, Flickr, YouTube, Patterns of production/consumption of information are relatively unexplored Usage patterns inform system design –Recommendation –Content pre-fetching –Spam detection

Where to go from here? Get real experience building systems! Opportunities: 496 projects –More projects: 491r/571r class next term –Autonomic Systems – USRA funding for summer –(applications due January 2010)