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 2013 Status Report
Summary:
A good quarter for publications, with 10 papers published (including 4 papers at HPCA’13) and 9 others accepted for publication. Satya (CMU), Babu Pillai (Intel Labs) and the cloudlets team had three papers accepted and another appearing, on using cloudlets for crowd sensing, video upload, and client offload. Guy Blelloch’s (CMU) paper on the Ligra graph processing framework appeared in PPoPP’13. New open source code is released, one release that builds on Mesos and three releases from the FAWN team. Mor Harchol-Balter’s (CMU) textbook entitled “Performance Analysis and Design of Computer Systems,” was published by Cambridge University Press, 2013 (http://www.amazon.com/Performance-Modeling-Design-Computer-Systems/dp/1107027500/). Awards and honors included:
- Onur Mutlu (CMU) has been appointed the Dr. William D. and Nancy W. Strecker Early Career Professor.
- David Andersen (CMU) and Michael Kaminsky (Intel) were awarded the Allen Newell Award for Research Excellence, given annually by the School of Computer Science at Carnegie Mellon University, for their work on the FAWN project.
- Three ISTC-CC funded PhD students from CMU were awarded prestigious PhD Fellowships: Aapo Kyrola (VMWare Fellowship), Gennady Pekhimenko (Microsoft Fellowship) and Julian Shun (Facebook Fellowship).
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
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).
Technical highlights
- Publications summary for the past quarter:
- 9 papers accepted: to HotOS'13 (2 papers), HPDC'13, IC2E'13, Mobisys'13 (2 papers), NSDI '13, SEA'13, and SIGMETRICS'13
- 10 papers published: in Alenex'13, HotMobile'13, HPCA'13 (4 papers), PPoPP'13, ACM TACO, IEEE TPDS, and VMWare Technical Journal.
- Open Source code releasesthis past quarter included:
- [Mesos] AirBnB recently released Chronos as an open source project. It is built on top of Mesos. See http://nerds.airbnb.com/introducing-chronos
- [FAWN] The FAWN team made three code releases this month:
- MemC3 (libcuckoo): https://github.com/efficient/libcuckoo
- Eiger:https://github.com/wlloyd/eiger
- NVMalloc/async memprotect: https://github.com/efficient/nvram
- [Ligra: graph processing framework] Guy Blelloch (CMU) and Julian Shun (CMU PhD student)’s paper entitled “Ligra: A Lightweight Graph Processing Framework for Shared-Memory,” appeared in PPoPP’13. Ligra has two very simple routines, one for mapping over edges and one for mapping over vertices. The routines can be applied to any subset of the vertices, which makes the framework useful for many graph traversal algorithms that operate on subsets of the vertices. Many graph algorithms expressed using this framework are very simple, concise, fast, and highly scalable.
- [Cloudlets] The Cloudlet team, Mahadev Satyanarayanan (CMU), Babu Pillai (Intel Labs), and CMU PhD student co-author, had a great quarter in terms of publications. First, a paper describing a new architecture for mobile crowd sensing appeared in HotMobile 2013. The key idea of this work is to improve scalability by reducing barriers to participation in crowd sensing. The system uses VMs on cloudlets to handle device/platform diversity, avoid the need to install a new app for each crowd sensing activity, overcome artificial sensor sharing limits, and automate participation based on user preferences. Second, a paper accepted to IC2E’13 investigates the performance of bandwidth and latency sensitive applications on local and remote cloud services, and demonstrates the case for a multi-tier cloud architecture with a largely unmanaged second tier (i.e., cloudlets). Third, a paper accepted to MobiSys’13 presents Gigasight, which improves scalability and feasibility of widespread video upload by utilizing near-edge cloud resources (cloudlets) to host and process streamed video data. This overcomes the WAN and datacenter ingress bottlenecks of a centralized solution, yet provides a means of indexing and searching videos. Finally, another paper accepted to Mobisys’13 describes and evaluates techniques to significantly reduce the time required to upload and execute a customized VM to nearby cloud resources—often to under 10 seconds from the start of offload to the first response to the client. This work significantly neutralizes one of the key arguments against VM-based offload—that it takes too long to transfer large VM images.
- [FAWN] The Fawn team, David Andersen (CMU), Michael Kaminsky (Intel Labs), and CMU PhD students, had a paper accepted to the 12th International Symposium on Experimental Algorithms (SEA 2013) describing more efficient version of the Rank and Select data structure commonly used as building blocks in succinct data structures. Another paper was accepted to HotOS’13 about how to build a lookup table with hundreds of billions of entries in DRAM on a commodity server. Finally, the team, including Michael Freedman (Princeton) and Wyatt Lloyd (Princeton PhD student), had four posters accepted to NSDI’13, with various subsets of authors, to go along with two papers accepted earlier to NSDI’13.
- Onur Mutlu (CMU) and co-authors had three papers appearing in HPCA’13. Margaret Martonosi (Princeton) also had an HPCA’13 paper.
Schedule of upcoming events and milestones
- March 28: ACM SOSP’13 submission deadline
- May 1: Joule Sort contest deadline
- June 17-21: ACM Sigmetrics’13 is being held at CMU
- June 25-26: Intel Symposium on Collaborative Research and Innovation, San Francisco, CA
- July 2: ACM SOCC’13 submission deadline.
Sponsor group interaction highlights
- Michael Kozuch (Intel Labs) and Mor Harchol-Balter (CMU) have been meeting with Edi Schmueli (Intel IDC) to determine if there are potential applications of the scheduling technology developed at CMU for the Intel NetBatch infrastructure.
- Working for Michael Kaminsky's (Intel Labs) organizing committee for the premier conference on operating systems, Garth Gibson (CMU) recruited commitments for $45,000 of sponsorship support for student travel grants.
- Garth Gibson (CMU) and Phil Gibbons (Intel Labs) met F2F with Michael Greene, Dani Napier Harrison, and Carl Rimby from Intel SSG, to discuss ISTC-CC’s research agenda.
- Greg Leeming (Intel) provided an additional 10 large-capacity SSDs to support ISTC-CC research.
- Jeff Parkhurst met with Gans on his collaborative work with OHSU and UCB. The work is aimed at identifying cancer risks and potential cures. He has two main focus areas he is looking for help. One is in the frameworks area(MapReduce, SparkGraph, MESOS, etc) where he is relying upon many tools from AMPLab. However, some of the work(such as MESOS) is already being funded through the ISTC CC. Additionally, Carlos Guestrin’s student that helped formulate the original GraphLab is a post doc student at UCB providing support. The other main focus area is machine learning which will come more into focus in the next few months.
- Justin visited the ISTC facility. Mike Kozuch gave an overview of the cloud computing center while Mei Chen gave an overview of the ISTC-EC automotive portion.
Other ISTC highlights
- Jeff Parkhurst submitted tutorial titled “Hitch Hikers Guide to Engaging in Intel Funded Academic Research and Programs” with Kimberly Sills and Susan Capello as co-authors to DTTC. We will hear back as to whether it was accepted on May 28.
- New IVY bridge samples sent to Margaret and Karsten as the old ones had issues. Thanks to Scott Buck for facilitating this.
- Aapo Kyrola (CMU PhD student) was awarded a 2013 Graduate Fellowship from the VMWare Academic Program to support his work on large-scale machine learning and graph computation.
- Gennady Pekhimenko (CMU PhD Student) was awarded a 2013 Microsoft Research PhD Fellowship to support his work on computer architecture.
- Julian Shun (CMU PhD student) was awarded a 2013 Facebook Fellowship to support his work on parallel computing.
- Phil Gibbons (Intel Labs), in his role as Editor-in-Chief, successfully launched the ACM Transactions on Parallel Computing, which is now accepting submissions. Phil also accepted an invitation to the Editorial Board of the newly launched IEEE Transactions on Cloud Computing.
- Michael Kaminsky (Intel Labs), in his role as General Chair, spent time this month working on SOSP’13 planning: a list of co-located conferences and workshops is now available: http://sigops.org/sosp/sosp13/workshops.html
- Garth Gibson (CMU) and Michael Kozuch (Intel Labs) served on the program committee for: 2013 ACM Measurement and Modeling of Computer Systems (SIGMETRICS). Mor Harchol-Balter (CMU) is the General Chair.
- Garth Gibson (CMU) agreed to participate in the technical program selection for: 2013 4th ACM Symposium on Cloud Computing (SOCC), San Jose, CA and 2014 USENIX Conference on File and Storage Technology (FAST), San Jose, CA.
- Phil Gibbons (Intel Labs), in his role as Sigmod’13 Area Chair for "Database Systems, Performance, Transaction Processing", ran the online program committee for the 21 papers submitted to that area.
- Michael Kozuch (Intel Labs) and Garth Gibson (CMU) attended the launch event for the Pittsburgh DataWorks (http://pghdataworks.org/), a not-for-profit effort to up-level the visibility of Big Data within Pittsburgh. 200+ attendees included folks from local/state government, industry, and academia. Contact Mike to suggest or to learn more about possible Intel engagements.
- Dave Andersen (CMU) wrote two blog posts advertising the upcoming NSDI papers:
http://da-data.blogspot.com/2013/03/optimistic-cuckoo-hashing-for.html
http://da-data.blogspot.com/2013/02/caring-about-causality-now-in-cassandra.html
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.]
- ''Resource-Aware Application State Monitoring", Shicong Meng, Srinivas R. Kashyap, Chitra Venkatramani and Ling Liu. IEEE Transactions on Parallel and Distributed Systems (TPDS’13), December 2012.
- "Elastic Resource Allocation in Datacenters: Gremlins in the Management Plane", Mukil Kesavan, Ada Gavrilovska, Karsten Schwan, VMware Technical Journal, December 2012.
- "TLB Improvements for CMPs: Inter-Core Cooperative Prefetchers and Shared Last-Level TLBs,” Dan Lustig, Abhishek Bhattacharjee, Margaret Martonosi, ACM Transactions on Architecture and Compiler Optimization (TACO’13), January, 2013.
- Hyeontaek Lim, David G. Andersen, Michael Kaminsky. “Practical batch-updatable external hashing with sorting.” In Workshop on Algorithm Engineering and Experiments (ALENEX'13), January 2013.
- "Reducing GPU Offload Latency Via Fine-Grained CPU-GPU Synchronization,” Dan Lustig and Margaret Martonosi, International Symposium on High-Performance Computer Architecture (HPCA’13), February, 2013.
- "MISE: Providing Performance Predictability and Improving Fairness in Shared Main Memory Systems," Lavanya Subramanian, Vivek Seshadri, Yoongu Kim, Ben Jaiyen, and Onur Mutlu, International Symposium on High-Performance Computer Architecture (HPCA’13), February, 2013.
- "Application-to-Core Mapping Policies to Reduce Memory System Interference in Multi-Core Systems," Reetuparna Das, Rachata Ausavarungnirun, Onur Mutlu, Akhilesh Kumar, and Mani Azimi, International Symposium on High-Performance Computer Architecture (HPCA’13), February, 2013.
- "Tiered-Latency DRAM: A Low Latency and Low Cost DRAM Architecture," Donghyuk Lee, Yoongu Kim, Vivek Seshadri, Jamie Liu, Lavanya Subramanian, and Onur Mutlu, International Symposium on High-Performance Computer Architecture (HPCA’13), February, 2013.
- “Lowering Barriers to Large-scale Mobile Crowdsensing,” Yu Xiao, Pieter Simoens, Padmanabhan Pillai, Kiryong Ha, and Mahadev Satyanarayanan, 14th ACM International Workshop on Mobile Computing Systems and Applications (HotMobile’13), February 2013.
- “Ligra: A Lightweight Graph Processing Framework for Shared-Memory,” Julian Shun and Guy Blelloch, 18th ACM SIGPLAN Symposium on Principles and Practice of Parallel Programming (PPoPP’13), February 2013.
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:
- Wyatt Lloyd (Princeton PhD Student) presented “Stronger Consistency and Semantics for Geo-Replicated Storage” at Georgia Tech on February 11, NEC Labs on February 19, Penn State on February 21, ATT Research on February 25, UC-Riverside on March 1, Univ. of Waterloo on March 8, USC on March 11, Univ. of Toronto on March 14 and at University College London on March 18. His talk is based on his ISTC-CC funded work.
- Kiryong Ha (CMU PhD Student) presented “The Impact of Mobile Applications on Data Center Consolidation” at the SDI/ISTC seminar series at CMU on March 21, 2013.
- Iulian Moraru, David Andersen, and Michael Kaminsky had their talk appear at the UCSD 2013 NVM Workshop. The talk covers the team’s previous work on NVMalloc and asynchronous mprotect. The work was well-received.