SEARCH
ISTC-CC NEWSLETTER
RESEARCH HIGHLIGHTS
Ling Liu's SC13 paper "Large Graph Processing Without the Overhead" featured by HPCwire.
ISTC-CC provides a listing of useful benchmarks for cloud computing.
Another list highlighting Open Source Software Releases.
Second GraphLab workshop should be even bigger than the first! GraphLab is a new programming framework for graph-style data analytics.
ISTC–CC November 2011 Status Report
Summary:
Tremendous amount of publications and presentations this month exhibiting the far reaching influence of the ISTC Cloud Computing research center. Specifically, 11 papers were published spanning 6 different conferences including 3 papers in SOSP’11 (the top computer systems conference), 4 papers in SOCC’11 (the top cloud computing conference). David Andersen’s (CMU) and Michael Kaminsky’s (Intel ISTC-CC) paper on SILT was the opening talk at SOSP’11, presenting a fast, memory-efficient key-value storage system for flash, which can scale to serve billions of key-value items on a single node. Ling Liu (GA Tech) gave a keynote on cloud computing at the Financial Services conference in Korea. The annual retreat of CMU’s Parallel Data Lab (PDL) included showcasing ISTC-CC research to 20 leading tech companies. GA Tech team visited Amazon AWS to better understand the challenges and opportunities faced 'at scale' by cloud providers. Plans for the first ISTC-CC retreat on Dec 8,9 are in full swing. Registration, hotel logistics, and agenda details are available at http://www.istc-cc.cmu.edu/events/retreat11.shtml.
Details:
ISTC Mission: Four inter-related research pillars (themes) architected to create a strong foundation for cloud computing of the future
The research agenda of the ISTC-CC is composed of the following four themes
- Specialization: Explores specialization as a primary means for order of magnitude improvements in efficiency (e.g., energy), including use of emerging technologies like non-volatile memory and specialized cores.
- Automation: Addresses cloud’s particular automation challenges, focusing on order of magnitude efficiency gains from smart resource allocation/scheduling and greatly improved problem diagnosis capabilities.
- Big Data: Addresses the critical need for cloud computing to extend beyond traditional big data usage (primarily, search) to efficiently and effectively support Big Data analytics, including the continuous ingest, integration, and exploitation of live data feeds (e.g., video or twitter).
- To the Edge: Explores new frameworks for edge/cloud cooperation that can efficiently and effectively exploit billions of context-aware clients and enable cloud-assisted client applications whose execution spans client devices, edge-local cloud resources, and core cloud resources.
Participants
Academic PI: Greg Ganger(CMU)
Executive Sponsor: Wen Hann Wang (CSR)
Managing Sponsor: Rich Uhlig (CSR-SAL)
Program Director: Jeff Parkhurst (APR)
Intel PI: Phil Gibbons
Intel Researchers:Michael Kiminsky, Mike Kozuch, Babu Pillai
Academic Partners: Dave Andersen, Guy Blelloch, Garth Gibson, Carlos Guestrin, Mor Harchol-Balter, Todd Mowry, Onur Mutlu, Priya Narasimhan, M. Satyanarayanan, Dan Siewiorek (CMU); Mike Freedman, Margaret Martonosi and Kai Li(Princeton); Randy Katz, Anthony Joseph, and Ion Stoica(UCB); Karsten Schwan, Ada Gavrilovska, Ling Liu, Carlton Pu, and Sudha Yalamanchili(G Tech).
ISTC Highlight Details:
- David Andersen’s (CMU) and Michael Kaminsky’s (Intel ISTC-CC) paper on SILT was the opening talk at SOSP’11. SILT (Small Index Large Table) is a memory-efficient, high-performance key-value store system based on flash storage that scales to serve billions of key-value items on a single node. Using a multi-store approach, entropy-coded tries, and partial-key cuckoo hashing, it requires only 0.7 bytes of DRAM per entry and retrieves key/value pairs using on average 1.01 flash reads each. See http://www.sigops.org/sosp/sosp11/current/2011-Cascais/printable/01-lim.pdf for the paper and https://github.com/silt/silt for the open source code.
- 3 papers from ISTC-CC researchers were published at the 23rd ACM Symposium on Operating Systems Principles (SOSP’11), the top computer systems conference: one by Randy Katz (UCB), and two by Dave Andersen (CMU) and Michael Kaminsky (Intel ISTC-CC), one of which was joint with Mike Freedman (Princeton).
- 4 papers from ISTC-CC researchers were published at the ACM Symposium on Cloud Computing (SOCC’11), the top cloud computing conference: one by Carlton Pu (GA Tech), one by Garth Gibson (CMU), and two by Dave Andersen (CMU) and Michael Kaminsky (Intel ISTC-CC), one of which was joint with Michael Kozuch (Intel ISTC-CC).
- Much of our ISTC-CC research was presented at the PDL Retreat, both as talks and posters (see “Presentations” below for a list of talks). The PDL Retreat was attended by 45 technical leaders from 20 companies, including Intel, Microsoft, Google, Facebook, VMware, EMC, HP, and Oracle. These attendees provide feedback on the research ideas, offer assistance (e.g., data center traces from Google and anecdotes from many), and help to promulgate the work by word of mouth and as technology transfer avenues.
- Ada Gavrilovska, Calton Pu, Karsten Schwan, and Matt Wolf (GA Tech) visited Amazon AWS on November 14, to better understand the challenges and opportunities faced 'at scale' by cloud providers, and to identify potential joint research projects with Amazon personnel.
- Intel’s Open Cirrus cluster here at the ISTC-CC continues to be heavily utilized, hitting the maximum this month so that computing resources became scarce. Disk failures remain a problem with 16 of the 40 blade machines currently running in single-disk state due to disk failures. Michael Kozuch and Michael Stroucken effected a number of improvements in the Open Cirrus infrastructure this month, including upgrading several of the racks to include 10 Gbps uplinks and implementing a new NAT for the cluster.
- The last step of the ISTC-CC launch completed this month with the receipt of funding by the four participating Universities. This funding was retroactive to September 1, so that student support started at the beginning of the semester. Thanks to the Universities for “interest-free loans” to cover the students until the funding started (and the loans were repaid).
Schedule of upcoming events and milestones
- First annual ISTC-CC Retreat, December 8-9, on CMU campus. All are welcome— Please see http://www.istc-cc.cmu.edu/events/retreat11.shtml to register and for the latest information on hotel reservations, travel logistics, and agenda.
- The next Open Cirrus Summit will be held in Beijing in June 2012. The CFP is posted at http://labs.chinamobile.com/cloud/opencirrus/OCsummit12, and the submission date is March 2. All are invited to submit short papers. Michael Kozuch is co-PC chair.
Technical highlights
- David Andersen’s (CMU) and Michael Kaminsky’s (Intel ISTC-CC) SILT paper published and open source code released (see above under ‘Further details on above Summary’).
- John Kelly (CMU grad student) was part of the team that did a 30 day trial using BCI (Brain Controlled Interaction) to control a prosthetic arm.
- Publications summary for the past month (partial list):
- 4 papers submitted: to Eurosys’12 (2 papers), PLDI’12, and ICDCS’12.
- 3 papers accepted: to ASPLOS’12 (2 papers), Journal of High Performance Computing Applications.
- 11 papers published: in SOSP’11 (3 papers), SOCC’11 (4 papers), HotPower’11, SC’11, IEEE Pervasive Computing, IEEE Engineering in Medicine and Biology Society Conference.
Sponsor group interaction highlights
- Frank Berry and Ted Willke from Intel Labs' SAL group attended the PDL Retreat in Bedford, PA.
- Babu Pillai (Intel ISTC-CC) introduced Rich Uhlig to Alex Hauptmann (CMU) and his work at the Informedia lab on algorithms for event detection in video streams. These algorithms have formed a basis of many of the interactive applications studied in SLIPstream and will be key for many of the cloud-edge and cloudlets research efforts.
- David Andersen (CMU) and Onur Mutlu (CMU) presented highlights of their cloud research to Wen-Hann Wang when he visited CMU for the ribbon-cutting event in late October.
- Michael Kaminsky met briefly with Ted Willke when he came to Pittsburgh for the Parallel Data Lab retreat. Michael showed Ted the current FAWN cluster and the new rack.
- Greg Ganger and Jeff Parkhurst met with Intel Fellow George Cox during his senior sponsor visit accompanied by Scott Buck and Carl Rimby. Greg and Jeff gave George an informal overview of the center and discussed avenues for technology transfer.
- Phil Gibbons and Jeff Parkhurst attended Intel SSG Fellow David Kuck’s presentation on co-design of HW/SW systems. We gave him an overview of the two ISTC centers at CMU.
Other ISTC highlights
- Ling Liu (Georgia Tech) is chairing the 'Cloud Computing' track for ICDCS 2012, and Calton Pu (Georgia Tech) is one of the PC members of that track.
- Michael Kozuch accepted the invitation to serve as co-PC chair for the next Open Cirrus Summit.
List of publications
[List of the publications that were PUBLISHED by center researchers during the month. This does not include submissions or acceptances, just publications.]
- [SOSP’11] “SILT: A Memory-Efficient, High-Performance Key-Value Store.” Hyeontaek Lim, Bin Fan, David G. Andersen, Michael Kaminsky. In 23nd ACM Symposium on Operating Systems Principles, Cascais, Portugal, October 2011.
- [SOSP’11] “Don't Settle for Eventual: Scalable Causal Consistency for Wide-Area Storage with COPS.” Wyatt Lloyd, Michael J. Freedman, Michael Kaminsky, David G. Andersen. In 23nd ACM Symposium on Operating Systems Principles, Cascais, Portugal, October 2011.
- [SOSP’11] “Design Implications for Enterprise Storage Systems via Multi-Dimensional Trace Analysis.” Y. Chen, K. Srinivasan, G. Goodson, R. Katz. In 23nd ACM Symposium on Operating Systems Principles, Cascais, Portugal, October 2011.
- [SOCC’11] “ActiveSLA: A Profit-Oriented Admission Control Framework for Database-as-a-Service Providers.” Pengcheng Xiong, Yun Chi, Shenghuo Zhu, Junichi Tatemura, Calton Pu, Hakan Hacigumus. In ACM Symposium on Cloud Computing, Cascais, Portugal, October 2011.
- [SOCC’11] “Small Cache, Big Effect: Provable Load Balancing for Randomly Partitioned Cluster Services.” Bin Fan, Hyeontaek Lim, David G. Andersen, Michael Kaminsky. In ACM Symposium on Cloud Computing, Cascais, Portugal, October 2011.
- [SOCC’11] “Switching the Optical Divide: Fundamental Challenges for Hybrid Electrical/Optical Datacenter Networks.” Hamid Hajabdolali Bazzaz, Malveeka Tewari, Guohui Wang, George Porter, T. S. Eugene Ng, David G. Andersen, Michael Kaminsky, Michael A. Kozuch, Amin Vahdat. In ACM Symposium on Cloud Computing, Cascais, Portugal, October 2011.
- [SOCC’11] “YCSB++: Benchmarking and Performance Debugging Advanced Features in Scalable Table Stores.” Swapnil Patil, Milo Polte, Kai Ren, Wittawat Tantisiriroj, Lin Xiao, Julio Lopez, Garth Gibson, Adam Fuchs, Billie Rinaldi. In ACM Symposium on Cloud Computing, Cascais, Portugal, October 2011.
- [HotPower’11] “The Case for Sleep States in Servers.” Anshul Gandhi, Mor Harchol-Balter, Michael Kozuch. In 4th ACM Workshop on Power-Aware Computing and Systems, Cascais, Portugal, October 2011.
- [SC’11] “Purlieus: Locality-aware Resource Allocation for MapReduce in a Cloud.” Balaji Palanisamy, Aameek Singh, Ling Liu, Bhushan Jain. In ACM/IEEE International Conference on High Performance Computing, Networking, Storage and Analysis, Seattle, WA, November 2011.
- “Adaptive Filter with Frequency Tracking and Variable Learning Rate for Line Noise Removal.” J. W. Kelly, J. L. Collinger, A. D. Degenhart, D. P. Siewiorek, A. Smailagic, W. Wang. Conference of the IEEE Engineering in Medicine and Biology Society, Boston, MA, September 2011.
- “User preferences for indicator and feedback modalities: A preliminary survey study for developing a coaching system to facilitate wheelchair power seat function usage.” H-Y. Liu, G. Grindle, F-C. Chuang, A. Kelleher, R. Cooper, D. Siewiorek, A. Smailagic, R. Cooper. IEEE Pervasive Computing, Vol. 10, October-December 2011.
List of presentations
- Ling Liu (Georgia Tech) gave a keynote on cloud computing at the Financial Services conference in Busan, Korea, November 2011.
- Hyeontaek Lim (CMU grad student) presented “SILT: A Memory-efficient High-performance Key-value Store” at SOSP’11, in Portugal, October 2011.
- Wyatt Lloyd (CMU grad student) presented “Don't Settle for Eventual: Scalable Causal Consistency for Wide-Area Storage with COPS” at SOSP’11, in Portugal, October 2011.
- Y. Chen (UCB grad student) presented “Design Implications for Enterprise Storage Systems via Multi-Dimensional Trace Analysis” at SOSP’11, in Portugal, October 2011.
- Pengcheng Xiong (GA Tech grad student) presented “ActiveSLA: A Profit-Oriented Admission Control Framework for Database-as-a-Service Providers” at SOCC’11, in Portugal, October 2011.
- Hamid Hajabdolali Bazzaz (CMU grad student) presented “Switching the Optical Divide: Fundamental Challenges for Hybrid Electrical/Optical Datacenter Networks” at SOCC’11, in Portugal, October 2011.
- Swapnil Patil (CMU grad student) presented “YCSB++: Benchmarking and Performance Debugging Advanced Features in Scalable Table Stores” at SOCC’11, in Portugal, October 2011.
- Anshul Gandhi (CMU grad student) presented "The Case for Sleep States in Servers" at HotPower’11, in Portugal, October 2011.
- Balaji Palanisamy (GA Tech grad student) presented “Purlieus: Locality-aware Resource Allocation for MapReduce in a Cloud” at SC’11, in Seattle, WA, November 2011.
- Tutorial "Heterogeneous Computing with GPU Ocelot", presented by Georgia Tech at the IEEE International Symposium on Workload Characterization, in Austin, TX, November 2011.
- ISTC-CC Talks from the 19th Annual PDL Workshop and Retreat, Bedford, PA
- Carlos Guestrin (CMU) presented “GraphLab: A New Parallel Framework for Machine Learning”
- Swapnil Patil (CMU grad student) presented “YCSB++: Benchmarking and Performance Debugging Advanced Features in Scalable Table Stores”
- Kai Ren (CMU grad student) presented “Resource Attribution and Metrics Correlation in Clouds”
- Iulian Moraru (CMU grad student) presented “Persistent, Protected and Cached: Building Blocks for Main Memory Data Stores”
- Jim Cipar (CMU grad student) presented “Jackrabbit: Improved Agility in Elastic Distributed Storage”
- Alexey Tumanov (CMU grad student) presented “Large-Scale Distributed Cluster Resource Scheduling”
- Raja Sambasivan (CMU grad student) presented “Diagnosing Performance Changes by Comparing Requests Flows”
- Mike Kasick (CMU grad student) presented “Black-Box Localization of Storage Problems in Parallel File Systems”
- Jiri Simsa (CMU grad student) presented “Efficient Exploratory Testing of Concurrent Systems”
- Hyeontaek Lim (CMU grad student) presented “SILT: A Memory-efficient High-performance Key-value Store”
- Wolf Richter (CMU grad student) presented “Virtual Contact Sheets”
- Lin Xiao (CMU grad student) presented “Scalable Metadata Service in HDFS”
- Elmer Garduno (CMU grad student) presented “Interactive User Interface for Diagnosis in MapReduce Systems”
- Elie Krevat (CMU grad student) presented “Managing Inter-Service Performance for Dependencies”
- Bin Fan (CMU grad student) presented “Provable Load Balancing for Randomly Partitioned Cluster Services”