About

Jeffrey S. Chase is a Professor and the Director of Graduate Studies in the Department of Computer Science at Duke University. His research area is operating systems and distributed systems, with a focus on cloud infrastructure services and federation, trust management and security in cloud environments, energy-aware computing, and automation for cloud-hosted applications. His work has included research on network storage, software-defined and end-system networking, and Internet service infrastructures. Recently, his interests have centered on issues of integrity and trust—specifically, where trust originates, how it can be represented and reasoned about in software, and how to build trustworthy software systems. Throughout his career, he has been involved in various activities including editorial work, program committees for major conferences, and leadership roles in initiatives such as the NSF GENI Control Framework Working Group and the ACM Symposium on Cloud Computing. He has advised numerous students at Duke, including doctoral and master's students, and has contributed extensively to the academic community through service and research.

Research topics

• Computer Science
• Database
• Computer Security
• Distributed computing
• Computer network
• Operating system
• World Wide Web
• Engineering
• Electrical engineering

Selected publications

• Enhancing distribution system resiliency using grid-forming fuel cell inverter

  2022 · 11 citations

  • Computer Science
  • Computer Science
  • Electrical engineering

  Legacy inverters interfacing distributed energy re-sources are traditionally grid-following (GFL) in nature. GFL assets typically follow real power and reactive power set points. Recently, inverters with grid-forming (GFM) capability are gaining attention because GFM assets can increase the resiliency of the distribution system under stressed conditions. These GFM inverters can use photovoltaics, batteries, or fuel cells as their energy source. In this paper, we present information on inverters interfacing fuel cell assets, specifically with GFM capability. By introducing a fuel cell-powered GFM coupled with hydrogen production and storage, the GFM can continuously provide GFM activities during periods of low renewable resource availability and/or during power outages exceeding typical electric battery duration. Finally, we present information on the need for updates to interconnection and interoperability standards that can be leveraged by utilities for including fuel cell inverters in their asset mixes.

  PublisherDOI

• Federated Authorization for Managed Data Sharing: Experiences from the ImPACT Project

  2021 · 2 citations

  1st authorCorresponding
  • Computer Science
  • Computer Security
  • Computer Science

  This paper presents the rationale and design of the trust plane for ImPACT, a federated platform for managed sharing of restricted data. Key elements of the architecture include Web-based notaries for credential establishment based on declarative templates for Data Usage Agreements, a federated authorization pipeline, integration of popular services for identity management, and programmable policy based on a logical trust model with a repository of linked certificates. We show how these elements of the trust plane work in concert, and set the ideas in context with principles of federated authorization. A focus and contribution of the paper is to explore limitations of the resulting architecture and tensions among competing design goals. We also point the way toward future extensions, including policy-checked data access from cloud-hosted data enclaves with enhanced defenses against data leakage and exfiltration.

  PublisherDOI

• Lightweight Inter-transaction Caching with Precise Clocks and Dynamic Self-invalidation

  arXiv (Cornell University) · 2020

  • Computer Science
  • Computer Science
  • Distributed computing

  Distributed, transactional storage systems scale by sharding data across servers. However, workload-induced hotspots result in contention, leading to higher abort rates and performance degradation. We present KAIROS, a transactional key-value storage system that leverages client-side inter-transaction caching and sharded transaction validation to balance the dynamic load and alleviate workload-induced hotspots in the system. KAIROS utilizes precise synchronized clocks to implement self-invalidating leases for cache consistency and avoids the overhead and potential hotspots due to maintaining sharing lists or sending invalidations. Experiments show that inter-transaction caching alone provides 2.35x the throughput of a baseline system with only intra-transaction caching; adding sharded validation further improves the throughput by a factor of 3.1 over baseline. We also show that lease-based caching can operate at a 30% higher scale while providing 1.46x the throughput of the state-of-the-art explicit invalidation-based caching.

  PublisherOA PDFDOI

• Session details: Up in the Clouds

  2020-06-22

  article1st authorCorresponding

  No abstract available.

  PublisherDOI

• Characterizing Output Bottlenecks of a Production Supercomputer

  ACM Transactions on Storage · 2019-11-30 · 11 citations

  articleOpen accessSenior author

  This article studies the I/O write behaviors of the Titan supercomputer and its Lustre parallel file stores under production load. The results can inform the design, deployment, and configuration of file systems along with the design of I/O software in the application, operating system, and adaptive I/O libraries. We propose a statistical benchmarking methodology to measure write performance across I/O configurations, hardware settings, and system conditions. Moreover, we introduce two relative measures to quantify the write-performance behaviors of hardware components under production load. In addition to designing experiments and benchmarking on Titan, we verify the experimental results on one real application and one real application I/O kernel, XGC and HACC IO, respectively. These two are representative and widely used to address the typical I/O behaviors of applications. In summary, we find that Titan’s I/O system is variable across the machine at fine time scales. This variability has two major implications. First, stragglers lessen the benefit of coupled I/O parallelism (striping). Peak median output bandwidths are obtained with parallel writes to many independent files, with no striping or write sharing of files across clients (compute nodes). I/O parallelism is most effective when the application—or its I/O libraries—distributes the I/O load so that each target stores files for multiple clients and each client writes files on multiple targets in a balanced way with minimal contention. Second, our results suggest that the potential benefit of dynamic adaptation is limited. In particular, it is not fruitful to attempt to identify “good locations” in the machine or in the file system: component performance is driven by transient load conditions and past performance is not a useful predictor of future performance. For example, we do not observe diurnal load patterns that are predictable.

  PublisherOA PDFDOI

• Multi-version Indexing in Flash-based Key-Value Stores

  arXiv (Cornell University) · 2019-12-02

  preprintOpen access

  Maintaining multiple versions of data is popular in key-value stores since it increases concurrency and improves performance. However, designing a multi-version key-value store entails several challenges, such as additional capacity for storing extra versions and an indexing mechanism for mapping versions of a key to their values. We present SkimpyFTL, a FTL-integrated multi-version key-value store that exploits the remap-on-write property of flash-based SSDs for multi-versioning and provides a tradeoff between memory capacity and lookup latency for indexing.

  PublisherOA PDFDOI

• Demand Response for Computing Centers

  2018-07-25 · 1 citations

  book-chapter1st authorCorresponding

  Systems 237 7.3 Platform Enhancements for Energy-Aware VM Management 2387.3.1 Coordinated VM Power Management with VirtualPower 239 7.3.1.1 VirtualPower Architectural Overview 239 7.3.1.2 Experimental Results 2447.3.2 Paravirtualized Management Interfaces for Platform Power Budgeting 247 7.3.2.1 QoS Feedback with Congestion Pricing 247 7.3.2.2 Experimental Results 2507.4 Power Management Mechanisms for Distributed Virtualized Platforms 253 7.4.1 System Managers for Distributed Power Budgeting 253 7.4.2 Coordinating Data Center Management with VPMTokens 255 7.4.2.1 Types of VPM Tokens 255 7.4.2.2 Managing Power with Budget Tokens 2577.4.3 Experimental Results 257 7.5 Related Work 2597.6 Conclusions and Future Challenges 261 References 2637.1 INTRODUCTION The semiconductor industry has shifted toward multicore system architectures to continue harnessing the resources made available by Moore’s law, while avoiding the power and thermal bottlenecks associated with highperformance single-processor architectures. The effects of this transition are already apparent in data center environments, where commodity servers often consist of dual-package configurations that allow on the order of ten processors to be provisioned on a single platform. The increasing density of computational elements is accompanied with pressure on other platform components, such as memory and I/O (input/output). The net result is the profileration of dense server platform and rack configurations. Although these systems offer significant advantages for scale-up workloads, many enterprise applications are not able to fully exploit such architectures. The reasons for this include the inability for the associated software to scale to a larger number of processors, as well as the varying loads exhibited by web service applications. Unmitigated, the inability for applications to consistently and effectively make use of multicore server platforms can result in stranded data center resources, which can equate to significant cost inefficiencies. What is needed, then, is a means to manage resources in a fluid fashion, thereby achieving an elastic data center where servers make up a fungible resource pool that can dynamically be provisioned to applications in an on-demand manner. It is becoming increasingly clear that virtualization technologies can help meet this goal.

  PublisherDOI

• The Future of Multi-Clouds: A Survey of Essential Architectural Elements

  2018-10-01 · 9 citations

  articleSenior author

  In this paper we present a vision of an environment composed of multiple independent cloud providers of various sizes, interconnected by programmable networks in which tenants may acquire resources from the providers and interconnect them together to serve a variety of distributed applications. Based on several years of working with GENI and NSF Cloud efforts in the US we present the essential elements of such an architecture and discuss their attributes. These elements perform functions that we claim are minimally necessary in order to realize a multi-cloud environment and include: provider and resource discovery services that rely on flexible, semantically rich description and query mechanisms, common meta-data services for maintaining information state of tenant resource allocations, and programmable interconnect mechanisms to create multi-provider Software-Defined Exchanges (SDXs) that allow tenants to control their connectivity using a declarative authorization logic.

  PublisherDOI

• Rethinking Security in the Era of Cloud Computing

  IEEE Security & Privacy · 2017-01-01 · 8 citations

  article

  Cloud computing has emerged as a dominant computing platform for the foreseeable future, resulting in an ongoing disruption to the way we build and deploy software. This disruption offers a rare opportunity to integrate new approaches to computer security. The aggregating effect of cloud computing and the role of cloud providers as trust anchors can significantly benefit computing security.

  PublisherDOI

• Predicting Output Performance of a Petascale Supercomputer

  2017-06-23 · 49 citations

  articleOpen access

  In this paper, we develop a predictive model useful for output performance prediction of supercomputer file systems under production load. Our target environment is Titan---the 3rd fastest supercomputer in the world---and its Lustre-based multi-stage write path. We observe from Titan that although output performance is highly variable at small time scales, the mean performance is stable and consistent over typical application run times. Moreover, we find that output performance is non-linearly related to its correlated parameters due to interference and saturation on individual stages on the path. These observations enable us to build a predictive model of expected write times of output patterns and I/O configurations, using feature transformations to capture non-linear relationships. We identify the candidate features based on the structure of the Lustre/Titan write path, and use feature transformation functions to produce a model space with 135,000 candidate models. By searching for the minimal mean square error in this space we identify a good model and show that it is effective.

  PublisherOA PDFDOI

Recent grants

• TWC: Frontier: Collaborative: Rethinking Security in the Era of Cloud Computing

  NSF · $900k · 2013–2019

Frequent coauthors

• Kosrow Nowroozi

  Rutgers, The State University of New Jersey

  45 shared

• Jerome H. Check

  Cooper Institute for Reproductive Hormonal Disorders

  34 shared

• Ahmad Nazari

  Tehran University of Medical Sciences

  28 shared

• J.H. Check

  Cooper Medical School of Rowan University

  25 shared

• Milind Vaze

  21 shared

• Henry M. Levy

  Google (United States)

  17 shared

• Amin Vahdat

  Google (United States)

  15 shared

• Sharad G. Joshi

  12 shared