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 Q1 2014 Status Report
Summary:
Leadership: Satya (CMU) and co-authors win Best Paper Award at IC2E’14 and Best Demo award at HotMobile’14. Onur Mutlu (CMU) and co-authors’ HPCA’14 paper appears in Best Paper Session. Greg Ganger (CMU) wins CMU’s Steven J. Fenves Award for Systems Research. Other leadership highlights include program chairing, new courses and degree programs, book chapters, amplifying funding, 12 papers accepted, and 13 papers published. See below for details.
Transferred: Open source code releases of Tachyon, Spark, MICA, and Libcuckoo.
Collaborations: SSG-STO-DCST Big Data Technology group is using UC Berkeley’s Spark/Shark/Tachyon and contributing to its open source code base. ISG, SMG, and IL/UXR are collaborating with ISTC-CC on Cars and Cloudlets. DCG is collaborating with Princeton on hetero memory consistency. Two papers in HPCA are jointly co-authored by IL/ADR and CMU researchers. Other discussions and collaborations are ongoing with DCG, DSD, IT, and IL/SSR.
Details:
Upcoming Events
- Carnegie Mellon University’s Parallel Data Lab (PDL) Spring Visit Day in Pittsburgh, PA, May 13.
Intel employees are welcome to attend. Contact Michael Kozuch, michael.a.kozuch@intel.com.
LEADERSHIP
Awards/Honors:
- Best Paper Award: “Agentless Cloud-wide Streaming of Guest File System Updates,” Wolfgang Richter, Canturk Isci, Jan Harkes, Benjamin Gilbert, Vasanth Bala, and Mahadev Satyanarayanan, The Second IEEE Conference on Cloud Engineering (IC2E’14).
- Best Paper Session: “Improving Cache Performance Using Read-Write Partitioning,” Samira Khan, Alaa Alameldeen, Chris Wilkerson, Onur Mutlu, and Daniel Jimenez, Proceedings of the 20th International Symposium on High- Computer Architecture (HPCA’14).
- Best Demo: “QuiltView: Glass-Sourced Video for Google Maps Queries,” Zhuo Chen, Wenlu Hu, Kiryong Ha, Jan Harkes, Benjamin Gilbert, Jason Hong, Asim Smailagic, Dan Siewiorek, and Mahadev Satyanarayanan, The 15th International Workshop on Mobile Computing Systems and Applications (HotMobile’14).
- Keynote: Mor Harchol-Balter (CMU) was invited to give Keynote talk at GreenMETRICS Workshop, affiliated with the SIGMETRICS Conference, to take place in June 2014.
- Greg Ganger (CMU) was awarded the 2014 Steven J. Fenves Award for Systems Research. The award is presented annually by CMU’s College of Engineering for contributions to systems research. Ganger is being recognized for his significant contributions to computer systems, in particular for his work on soft updates and self-storage systems.
Conference Leadership:
- Karsten Schwan (GA Tech) and Vanish Talwar (HP) are again organizing the MBDS (Management of Big Data Systems) for the annual ICAC conference - paper submissions welcome.
- Karsten Schwan (GA Tech) was PC co-chair for ACM Middleware in Beijing, December 2013.
- Phil Gibbons (IL) served as chair for the "Big Data, Data Management and Analytics" track of the 34th International Conference on Distributed Computing Systems (ICDCS'14). Garth Gibson (CMU), Michael Kozuch (IL), Ling Liu (GA Tech), Priya Narasimhan (CMU), Calton Pu (GA Tech), and Karsten Schwan (GA Tech) served on the ICDCS’14 PC.
- Mor Harchol-Balter (CMU) joined 3 technical program committees: ACM SIGMETRICS, IFIP PERFORMANCE, and ECQT (European Conference on Queueing Theory).
- Phil Gibbons (IL) joined the PC for SOCC’14, and reviewed papers as a PC member for SPAA'14.
Books/Curriculum:
- Mor Harchol-Balter’s (CMU) textbook, "Performance Modeling & Design of Computer Systems" is being used at Columbia University and Washington University, St. Louis in this semester's classes.
- Yoongu Kim (CMU grad student) and Onur Mutlu (CMU) wrote an invited book chapter, "Memory Systems," for The Computing Handbook, Third Edition: Computer Science and Software Engineering, CRC Press, to be published in April 2014.
- First offering of 15-719, Advanced Cloud Computing, www.cs.cmu.edu/~15719 at CMU. 51 students enrolled in Fall 2013. Co-instructors: Greg Ganger, Majd Sakr, Raja Sambasivan. Supported by Intel education development for 2013-2014.
- Program development and graduation (12/2013) of first class (5) of Systems Track in the Masters of Computational Data Science, mcds.cs.cmu.edu, at CMU. Admission (9/2013) of second class (18). Program involves graduate projects courses in distributed systems, cloud computing, storage systems, computer architecture and parallel programming, operating systems, database systems, data mining and machine learning. Garth Gibson (CMU) is the co-director.
Amplifying Funding:
- Karsten Schwan, Ada Gavrilovska, Greg Eisenhauer, and Matt Wolf (GA Tech), "Hobbes: OS and Runtime Support for Application Composition'', Department of Energy, ASCR, $125,000/yr, Sept. 2013 - Aug. 2016, funding received December 2013.
- Ada Gavrilovska (GA Tech) “Heterogeneous Machines,” HP Corporation, $25,000, November 2013, and $50,000, February 2014.
Publications summary for the past quarter:
- 12 papers accepted: to EuroSys'14 (2 papers), ISCA (2 papers), SIGMETRICS (2 papers), CCGrid'14, DSN'14, JACM, MobiSys'14, Probability in the Engineering and Informational Sciences, and SIGMOD'14.
- 13 papers published: in HPCA'14 (4 papers), FAST'14 (2 papers), IEEE Trans. on Service Computing (2 papers), ASPLOS'14, CGO'14, GPGPU-7, HotMobile'14, and IC2E'14.
Transferred
Open Source code releases this past quarter included:
- Berkeley had three releases:
- Tachyon 0.4.1 (http://tachyon-project.org/)
- Spark 0.9.0 (https://spark.incubator.apache.org/releases/spark-release-0-9-0.html, Feb 2, 2014)
- Spark 0.8.1 (https://spark.incubator.apache.org/news/spark-0-8-1-released.html, Dec 19, 2013)
Note: Intel SSG/STO/DCST - Big Data Technology group is using Spark/Shark/Tachyon and contributing to its open source code base.
- CMU had two releases:
- Networked key-value cache MICA ( https://github.com/efficient/mica)
- Concurrent cuckoo hash table (https://github.com/efficient/libcuckoo)
Intel / ISTC-CC Collaborations
- [IL/CMU] M. Satyanarayanan (CMU) organized a successful F2F discussion on cloudlets in cars and in cellular base stations, which included representatives from Vodaphone, Bosch Research, IBM, and Intel (Babu Pillai (IL), Phil Gibbons (IL), Ren Wang (IL/SSR), and Jack Weast (IL/UXR)). Key discussions covered the research directions and interests in the cloudlets space of the different companies, and potential avenues of collaborations, such as getting research access to cellular base station HW or a set of actual application VMs for improved benchmarking and evaluation of software stacks.
- [ISG/SMG/IL] Babu Pillai (IL) has continued to actively participate in the Cars and Cloudlets effort with Jack Weast (IL/UXR), Ryan Brotman (IL/UXR), Robert Adams (IL/UXR), George Moakley (ISG), Hugh Mercer (SMG), and Ren Wang (IL/SSR). He also participated in a F2F brainstorming session to come up with use case scenarios.
- [IL/CMU] David Andersen (CMU), Michael Kaminsky (IL), and Dong Zhou (CMU grad student) continue to work with Ren Wang (IL/SSR) about the cluster version of their Cuckooswitch work.
- [DCG/Princeton] Margaret Martonosi (Princeton) and Dan Lustig (Intel Graduate Fellow, Princeton grad student) continued their collaboration with Intel’s Michael Pellauer (DCG) regarding specification languages for memory consistency models that can assist in the design, verification, and programming of heterogeneous systems. In particular, this work addresses the significant challenges inherent in building systems where processor cores, accelerators, and other heterogeneous compute elements access memory with different ordering assumptions. First-year Princeton PhD student (and recent Intel intern) Caroline Trippel has recently joined the project.
- [DCG/IL] Michael Kozuch (IL) discussed ISTC research results with many Intel Labs researchers during his visit to Portland on Feb 26-27. Michael also met with Ran Kurup (DCG) to discuss university research.
- [IT/CMU] Jeff Sedayao (IT) is testing out EPaxos code from Iulian Moreau (CMU grad student), David Andersen (CMU), and Michael Kaminsky (IL).
- [IL] Phil Gibbons and Patrick Stolt (IL/SPC) discussed ISTC-CC efforts in Big Data and future trends in Big Data and HPC. Pat is pulling together a research portfolio for IL on HPC and Big Data.
- [DSD/IL] Phil Gibbons (IL), Jeff Parkhurst (IL), and Ted Willke (DSD) discussed potential and ongoing ISTC-CC/DSD collaborations. As a next step, Phil will present an ISTC-CC overview at Ted’s staff on March 20.
- [IL/CMU] Two papers in HPCA’14 were co-authored by IL/ADR and CMU: “Improving Cache Performance Using Read-Write Partitioning,” by Samira Khan, Alaa Alameldeen, Chris Wilkerson, Onur Mutlu, and Daniel Jimenez and “Improving DRAM Performance by Parallelizing Refreshes with Accesses,” by Kevin Chang, Donghyuk Lee, Zeshan Chishti, Chris Wilkerson, Alaa Alameldeen, Yoongu Kim, and Onur Mutlu.
- [SSG/STO/DCST/Berkeley] Intel’s SSG/STO/DCST Big Data Technology group is using Spark/Shark/Tachyon and contributing to its open source code base.
Proof of Concept
- [Cloudlets and wearables] The cloudlets team has developed a prototype system that leverages compute capacity of cloudlets and first-person sensing of Google Glass-like devices to provide extensible, wearable cognitive assistance. The system is intended to help those in cognitive decline by reading signs, identifying faces, providing reminders and social cues, etc. The VM-based architecture permits multiple open source and commercial computer-vision-based services running on both Windows and Linux environments to be used concurrently. A paper by Kiryong Ha, Zhuo Chen, Wenlu Hu, Wolfgang Richter (CMU grad students), Babu Pillai (IL) and M. Satyanarayanan (CMU) has been accepted to MobiSys 2014.
- [SpringFS] Elastic storage systems can be expanded or contracted to meet current demand, allowing servers to be turned off or used for other tasks. However, the usefulness of an elastic distributed storage system is limited by its agility: how quickly it can increase or decrease its number of servers. We developed SpringFS, an elastic storage system that can quickly change its number of active servers, while retaining elasticity and performance goals. SpringFS uses several novel techniques to minimize the work needed before deactivation or activation of servers. Analysis of real-world traces from Hadoop deployments at Facebook and various Cloudera customers and experiments with the SpringFS prototype confirm SpringFS's agility, show that it reduces the amount of data migrated for elastic resizing by up to two orders of magnitude, and show that it cuts the percentage of active servers required by 67–82%, outdoing state-of-the-art designs by 6–120%. A paper by Lianghong Xu, James Cipar, Elie Krevat, Alexey Tumanov, Nitin Gupta (CMU grad students), Michael Kozuch (IL), and Greg Ganger (CMU) appeared in FAST’14.
List of publications
- “Who Is Your Neighbor: Net I/O Performance Interference in Virtualized Clouds,” Xing Pu, Ling Liu, Yiduo Mei, Sankaran Sivathanu, Younggyun Koh, Calton Pu and Yuanda Cao, IEEE Transactions on Services Computing, 2013.
- “ADAM: Genomics Formats and Processing Patterns for Cloud Scale Computing,” Matt Massie, Frank Nothaft, Christopher Hartl, Christos Kozanitis, Andre Schumacher, Anthony D. Joseph, and David A. Patterson, Berkeley Technical Report, December 2013.
- “GraphX: Unifying Data-Parallel and Graph-Parallel Analytics,” Reynold Xin, Dan Crankshaw, Ankur Dave, Joseph Gonzalez, Michael Franklin, and Ion Stoica, ArXiv, February 2014.
- “Red Fox: An Execution Environment for Accelerating Relational Queries using GPUs,” H. Wu, G. Diamos, T. Sheard, M. Aref, and S. Yalamanchili, IEEE/ACM International Symposium on Code Generation and Optimization (CGO’14), February 2014.
- “Reducing the Cost of Persistence for Nonvolatile Heaps in End User Devices,” Sudarsun Kannan, Ada Gavrilovska, and Karsten Schwan, Proceedings of the 20th International Symposium on High-Performance Computer Architecture (HPCA’14), February 2014.
- “Improving Cache Performance Using Read-Write Partitioning,” Samira Khan, Alaa Alameldeen, Chris Wilkerson, Onur Mutlu, and Daniel Jimenez, Proceedings of the 20th International Symposium on High- Computer Architecture (HPCA’14), February 2014.
- “Improving DRAM Performance by Parallelizing Refreshes with Accesses,” Kevin Chang, Donghyuk Lee, Zeshan Chishti, Chris Wilkerson, Alaa Alameldeen, Yoongu Kim, and Onur Mutlu, Proceedings of the 20th International Symposium on High-Performance Computer Architecture (HPCA’14), February 2014.
- “MRPB: Memory Request Prioritization for Massively Parallel Processors,” Wenhao Jia, Kelly A. Shaw, and Margaret Martonosi, Proceedings of the 20th International Symposium on High-Performance Computer Architecture (HPCA’14), February 2014.
- “QuiltView: Glass-Sourced Video for Google Maps Queries,” Zhuo Chen, Wenlu Hu, Kiryong Ha, Jan Harkes, Benjamin Gilbert, Jason Hong, Asim Smailagic, Dan Siewiorek, and Mahadev Satyanarayanan, The 15th International Workshop on Mobile Computing Systems and Applications (HotMobile’14), February 2014.
- “SpringFS: Bridging Agility and Performance in Elastic Distributed Storage,” by Lianghong Xu, James Cipar, Elie Krevat, Alexey Tumanov, Nitin Gupta, Michael A. Kozuch, and Gregory R. Ganger, the 12th USENIX Conference on File and Storage Technologies (FAST’14), February 2014.
- “Toward Strong, Usable Access Control for Shared Distributed Data,” Michelle L. Mazurek, Yuan Liang, William Melicher, Manya Sleeper, Lujo Bauer, Gregory R. Ganger, Nitin Gupta, and Michael K. Reiter, the 12th USENIX Conference on File and Storage Technologies (FAST’14), February 2014.
- “Probabilistic Diffusion of Social Influence with Incentives,” Myungcheol Doo, and Ling Liu, Special Issue on Clouds for Social Computing, IEEE Transactions on Service Computing.
- “Guardrail: A High Fidelity Approach to Protecting Hardware Devices from Buggy Drivers,” Olatunji Ruwase, Phillip B. Gibbons, Michael A. Kozuch, and Todd Mowry, 19th International Conference on Architectural Support for Programming Languages and Operating Systems (ASPLOS’14), March 2014.
- “ParallelJS: An Execution Framework for Javascript on Heterogeneous Systems,” J. Wang, N. Rubin, and S. Yalamanchili, Proceedings of Seventh Workshop on General-Purpose Computation on Graphics Processing Units (GPGPU-7), March 2014.
- “Agentless Cloud-wide Streaming of Guest File System Updates,” Wolfgang Richter, Canturk Isci, Jan Harkes, Benjamin Gilbert, Vasanth Bala, and Mahadev Satyanarayanan, The Second IEEE Conference on Cloud Engineering (IC2E’14), March 2014.
List of presentations
In addition to the conference/workshop presentations associated with each of the published conference/workshop papers listed above, we had the following presentations:
- Mor Harchol-Balter (CMU) gave an invited talk on “Dynamic Power Management of Data Centers: Theory and Practice” at DIMACS Working Group on Algorithms for Green Data Storage, Dec 2013
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
AcademicPartners: Dave Andersen, Guy Blelloch, Garth Gibson, Carlos Guestrin, Mor Harchol-Balter, Todd Mowry, Onur Mutlu, Priya Narasimhan, M. Satyanarayanan, and Dan Siewiorek (CMU); Mike Freedman, Kai Li, and Margaret Martonosi (Princeton); Anthony Joseph, Randy Katz, and Ion Stoica (UC Berkeley); Ada Gavrilovska, Ling Liu, Calton Pu, Karsten Schwan, and Sudha Yalamanchili (GA Tech).