| NSF Org: |
CNS Division Of Computer and Network Systems |
| Recipient: |
|
| Initial Amendment Date: | June 24, 2013 |
| Latest Amendment Date: | June 24, 2013 |
| Award Number: | 1319405 |
| Award Instrument: | Standard Grant |
| Program Manager: |
Marilyn McClure
mmcclure@nsf.gov (703)292-5197 CNS Division Of Computer and Network Systems CSE Directorate for Computer and Information Science and Engineering |
| Start Date: | September 1, 2013 |
| End Date: | August 31, 2017 (Estimated) |
| Total Intended Award Amount: | $499,413.00 |
| Total Awarded Amount to Date: | $499,413.00 |
| Funds Obligated to Date: |
|
| History of Investigator: |
|
| Recipient Sponsored Research Office: |
21 N PARK ST STE 6301 MADISON WI US 53715-1218 (608)262-3822 |
| Sponsor Congressional District: |
|
| Primary Place of Performance: |
21 North Park Street Madison WI US 53715-1218 |
| Primary Place of
Performance Congressional District: |
|
| Unique Entity Identifier (UEI): |
|
| Parent UEI: |
|
| NSF Program(s): | CSR-Computer Systems Research |
| Primary Program Source: |
|
| Program Reference Code(s): |
|
| Program Element Code(s): |
|
| Award Agency Code: | 4900 |
| Fund Agency Code: | 4900 |
| Assistance Listing Number(s): | 47.070 |
ABSTRACT
Research and development in cloud-based distributed systems is currently
hindered due to the incredible difficulty of running such systems at
scale. The Wisconsin Pocket Datacenter (PoD) project directly addresses this
dilemma by enabling researchers and developers to run large-scale distributed
systems on modest hardware resources. The key novel idea behind PoD is data
elision, which removes actual application data from the underlying disks,
memories, and network transfers of the distributed system under test. PoD thus
enables many virtual instances of such a system to be run upon just a few
physical machines, thus making testing, debugging, and in general development
simpler, faster, and less resource-intensive.
Cloud-based distributed systems represent a critically-important piece of
modern computing infrastructure. As users and corporations migrate their data
and computation into such facilities, it is increasingly important to ensure
that such systems operate correctly and efficiently. PoD enables developers to
debug existing systems and test new designs without excessive hardware
resources; with such a tool in their workbench, system developers will be able
to more readily build and test the next generation of cloud-based systems to
support the needs of a growing percentage of society.
PROJECT OUTCOMES REPORT
Disclaimer
This Project Outcomes Report for the General Public is displayed verbatim as submitted by the Principal Investigator (PI) for this award. Any opinions, findings, and conclusions or recommendations expressed in this Report are those of the PI and do not necessarily reflect the views of the National Science Foundation; NSF has not approved or endorsed its content.
Large-scale storage systems are central to the modern information economy. Behind every important Internet service, including Google's Gmail, Facebook's social network, etc., lies the data that drives each service; thus, the scalable storage systems that store the vast quantities of data inside each service, and specifically their performance and reliability characteristics, are of extreme importance.
The "Pocket Datacenter" Project at the University of Wisconsin-Madison addressed these central problems in the design and implementation of large-scale computer systems. How can researchers and practitioners analyze existing large-scale storage systems so as to better understand their deficiencies? How can these results help drive the creation of new and improved storage systems?
The research done within the confines of the project advanced the state of the art in systems evaluation techniques in numerous critical directions. For example, the Chopper tool was developed and can pinpoint high tail latencies in file systems, using advanced statistical methods based on Latin Hypercube sampling. In addition, a new approach to reducing the high costs of translation in modern SSD-based storage systems was developed with the File System Devirtualizer (FSDV). A third example is the extension of previous emulation technology; although originally developed to analyze storage systems, here the basic idea was applied to improve storage performance in virtualized environments with a technique called System-Call Interception (SCI). A fourth advancement came in the form of a tool to analyze the robustness of existing distributed storage systems to correlated failures called the Protocol-Aware Crash Explorer (PACE). Finally, an extensive SSD simulation environment called WiscSim was created, enabling the deep analysis of the modern storage systems stack, finding many flaws in existing designs as well as suggesting new directions.
Overall, the Pocket Datacenter project has advanced our knowledge of how high-end storage systems should be designed, as well as our understanding of how to build tools to perform such analyses. The end goal of helping to realize significant improvements in existing systems, and helping to shape the future design and implementation of such systems, was thus achieved.
Last Modified: 11/29/2017
Modified by: Remzi H Arpaci-Dusseau
Please report errors in award information by writing to: awardsearch@nsf.gov.
