Session Chairs: Gennady Pekhimenko, University of Toronto / Vector Institute, and Shivaram Venkataraman, University of WisconsinMadison, Aurick Qiao, Petuum, Inc. and Carnegie Mellon University; Sang Keun Choe and Suhas Jayaram Subramanya, Carnegie Mellon University; Willie Neiswanger, Petuum, Inc. and Carnegie Mellon University; Qirong Ho, Petuum, Inc.; Hao Zhang, Petuum, Inc. and UC Berkeley; Gregory R. Ganger, Carnegie Mellon University; Eric P. Xing, MBZUAI, Petuum, Inc., and Carnegie Mellon University. Uniquely, Dorylus can take advantage of serverless computing to increase scalability at a low cost. Moreover, to handle dynamic workloads, Nap adopts a fast NAL switch mechanism. Using selective profiling, we build DMon, a system that can automatically locate data locality problems in production, identify access patterns that hurt locality, and repair such patterns using targeted optimizations. Abstract registrations that do not provide sufficient information to understand the topic and contribution (e.g., empty abstracts, placeholder abstracts, or trivial abstracts) will be rejected, thereby precluding paper submission. Authors are required to register abstracts by 3:00 p.m. PST on December 3, 2020, and to submit full papers by 3:00 p.m. PST on December 10, 2020. Conference site 49 papers accepted out of 251 submitted. When uploading your OSDI 2021 reviews for your submission to SOSP, you can optionally append a note about how you addressed the reviews and comments. Researchers from the Software Systems Laboratory bagged Best Paper Awards at the 15th USENIX Symposium on Operating Systems Design and Implementation (OSDI 2021) and the 2021 USENIX Annual Technical Conference (USENIX ATC 2021).. Jay Lepreau Best Paper Award, OSDI'21. Please identify yourself as a presenter and include your mailing address in your email. For example, talks may be shorter than in prior years, or some parts of the conference may be multi-tracked. (Visa applications can take at least 30 working days to process.) Academic and industrial participants present research and experience papers that cover the full range of theory and practice of computer . The conference papers and full proceedings are available to registered attendees now and will be available to everyone beginning Wednesday, July 14, 2021. As the emerging trend of graph-based deep learning, Graph Neural Networks (GNNs) excel for their capability to generate high-quality node feature vectors (embeddings). Starting with small invariant formulas and strongest possible invariants avoids large SMT queries, improving SMT solver performance. We compare Marius against two state-of-the-art industrial systems on a diverse array of benchmarks. Existing frameworks optimize tensor programs by applying fully equivalent transformations, which maintain equivalence on every element of output tensors. Password This paper presents Dorylus: a distributed system for training GNNs. We propose PET, the first DNN framework that optimizes tensor programs with partially equivalent transformations and automated corrections. Only two types of supplementary material are permitted: source code described in the paper and formal proofs sketched in the paper. Youngseok Yang, Seoul National University; Taesoo Kim, Georgia Institute of Technology; Byung-Gon Chun, Seoul National University and FriendliAI. Based on this observation, P3 proposes a new approach for distributed GNN training. The key insight guiding our design is computation separation. OSDI'21 accepted 31 papers and 26 papers participated in the AE, a significant increase in the participate ratio: 84%, compared to OSDI'20 (70%) and SOSP'19 (61%). However, a plethora of recent data breaches show that even widely trusted service providers can be compromised. Proceedings Front Matter USENIX new Date().getFullYear()>document.write(new Date().getFullYear()); Grants for Black Computer Science Students Application, Propose an interesting, compelling solution, Demonstrate the practicality and benefits of the solution, Clearly describe the paper's contributions, Clearly articulate the advances beyond previous work. OSDI takes a broad view of the systems area and solicits contributions from many fields of systems practice, including, but not limited to, operating systems, file and storage systems, distributed systems, cloud computing, mobile systems, secure and reliable systems, systems aspects of big data, embedded systems, virtualization, networking as it relates to operating systems, and management and troubleshooting of complex systems. Computation separation makes it possible to construct a deep, bounded-asynchronous pipeline where graph and tensor parallel tasks can fully overlap, effectively hiding the network latency incurred by Lambdas. With an aim to improve time-to-accuracy performance in model training, Oort prioritizes the use of those clients who have both data that offers the greatest utility in improving model accuracy and the capability to run training quickly. We first introduce two new hardware primitives: 1) Guarded Page Table (GPT), which protects page table pages to support page-level secure memory isolation; 2) Mountable Merkle Tree (MMT), which supports scalable integrity protection for secure memory. NrOS is primarily constructed as a simple, sequential kernel with no concurrency, making it easier to develop and reason about its correctness. Zeph executes privacy-adhering data transformations in real-time and scales to thousands of data sources, allowing it to support large-scale low-latency data stream analytics. Session Chairs: Nadav Amit, VMware Research Group, and Ada Gavrilovska, Georgia Institute of Technology, Stephen Ibanez, Alex Mallery, Serhat Arslan, and Theo Jepsen, Stanford University; Muhammad Shahbaz, Purdue University; Changhoon Kim and Nick McKeown, Stanford University. Foreshadow was chosen as an IEEE Micro Top Pick. (Oct 2018) Awarded an Intel Faculty Grant for Research on automated performance optimization (Sep. 2018) Our paper on Foreshadow is accepted to appear at USENIX Security. We have implemented a prototype of our design based on Penglai, an open-sourced enclave system for RISC-V. However, with the increasingly speedy transactions and queries thanks to large memory and fast interconnect, commodity HTAP systems have to make a tradeoff between data freshness and performance degradation. We introduce a hybrid cryptographic protocol for privacy-adhering transformations of encrypted data. We also propose two file system techniques for ZNS+-aware LFS. We demonstrate that Marius achieves the same level of accuracy but is up to one order of magnitude faster. AI enables principled representation of knowledge, complex strategy optimization, learning from data, and support to human decision making. Instead, we propose addressing the root cause of the heuristics problem by allowing software to explicitly specify to the device if submitted requests are latency-sensitive. P3 exposes a simple API that captures many different classes of GNN architectures for generality. Papers must be in PDF format and must be submitted via the submission form. PC members are not required to read supplementary material when reviewing the paper, so each paper should stand alone without it. Leveraging these information, Pollux dynamically (re-)assigns resources to improve cluster-wide goodput, while respecting fairness and continually optimizing each DL job to better utilize those resources. In this paper, we present P3, a system that focuses on scaling GNN model training to large real-world graphs in a distributed setting. Some recent schedulers choose job resources for users, but do so without awareness of how DL training can be re-optimized to better utilize the provided resources. We implement a variant of a log-structured merge tree in the storage device that not only indexes file objects, but also supports transactions and manages physical storage space. A.H. Hunter, Jane Street Capital; Chris Kennelly, Paul Turner, Darryl Gove, Tipp Moseley, and Parthasarathy Ranganathan, Google. However, memory allocation decisions also impact overall application performance via data placement, offering opportunities to improve fleetwide productivity by completing more units of application work using fewer hardware resources. We have made Fluffy publicly available at https://github.com/snuspl/fluffy to contribute to the security of Ethereum. However, your OSDI submission must use an anonymized name for your project or system that differs from any used in such contexts. Notification of conditional accept/reject for revisions: 3 March 2022. Distributed systems are notoriously hard to implement correctly due to non-determinism. blk-switch uses this insight to adapt techniques from the computer networking literature (e.g., multiple egress queues, prioritized processing of individual requests, load balancing, and switch scheduling) to the Linux kernel storage stack. Camera-ready submission (all accepted papers): 15 Mars 2022. In this talk, I'll speculate on how we came to this unfortunate state of affairs, and what might be done to fix it. Prior or concurrent workshop publication does not preclude publishing a related paper in OSDI. Responses should be limited to clarifying the submitted work. Nico Lehmann and Rose Kunkel, UC San Diego; Jordan Brown, Independent; Jean Yang, Akita Software; Niki Vazou, IMDEA Software Institute; Nadia Polikarpova, Deian Stefan, and Ranjit Jhala, UC San Diego. The ZNS+ also allows each zone to be overwritten with sparse sequential write requests, which enables the LFS to use threaded logging-based block reclamation instead of segment compaction. 64 papers accepted out of 341 submitted. Secure Computation (SC) is a family of cryptographic primitives for computing on encrypted data in single-party and multi-party settings. The 15th USENIX Symposium on Operating Systems Design and Implementation (OSDI '21) will take place as a virtual event on July 1416, 2021. A significant obstacle to using SC for practical applications is the memory overhead of the underlying cryptography. Consensus bugs are bugs that make Ethereum clients transition to incorrect blockchain states and fail to reach consensus with other clients. Further, Vegito can recover from cascading machine failures by using the columnar backup in less than 60 ms. This is unfortunate because good OS design has always been driven by the underlying hardware, and right now that hardware is almost unrecognizable from ten years ago, let alone from the 1960s when Unix was written. How can we design systems that will be reliable despite misbehaving participants? Swapnil Gandhi and Anand Padmanabha Iyer, Microsoft Research. Owing to the sequential write-only zone scheme of the ZNS, the log-structured file system (LFS) is required to access ZNS solid-state drives (SSDs). Authors are also encouraged to contact the program co-chairs, osdi21chairs@usenix.org, if needed to relate their OSDI submissions to relevant submissions of their own that are simultaneously under review or awaiting publication at other venues. We describe PrivateKube, an extension to the popular Kubernetes datacenter orchestrator that adds privacy as a new type of resource to be managed alongside other traditional compute resources, such as CPU, GPU, and memory. Used Zotero to organize papers about the stress and diffusion between anode and electrolyte and made a summary . All the times listed below are in Pacific Daylight Time (PDT). To remedy this, we introduce DeSearch, the first decentralized search engine that guarantees the integrity and privacy of search results for decentralized services and blockchain apps. The abstractions we design for the privacy resource mirror those defined by Kubernetes for traditional resources, but there are also major differences. For example, optimistic concurrency control (OCC) is better than two-phase-locking (2PL) under low contention, while the converse is true under high contention. Fan Lai, Xiangfeng Zhu, Harsha V. Madhyastha, and Mosharaf Chowdhury, University of Michigan. sosp ACM Symposium on Operating Systems Principles. Collaboration: You have a collaboration on a project, publication, grant proposal, program co-chairship, or editorship within the past two years (December 2018 through March 2021). Qing Wang, Youyou Lu, Junru Li, and Jiwu Shu, Tsinghua University. Academic and industrial participants present research and experience papers that cover the full range of theory . The 15th USENIX Symposium on Operating Systems Design and Implementation seeks to present innovative, exciting research in computer systems. Accepted papers will be allowed 14 pages in the proceedings, plus references. Novel system designs, thorough empirical work, well-motivated theoretical results, and new application areas are all . Federated Learning (FL) is an emerging direction in distributed machine learning (ML) that enables in-situ model training and testing on edge data. GoJournal is implemented in Go, and Perennial is implemented in the Coq proof assistant. OSDI brings together professionals from academic and industrial backgrounds in a premier forum for discussing the design, implementation, and implications of systems software. We identify that current systems for learning the embeddings of large-scale graphs are bottlenecked by data movement, which results in poor resource utilization and inefficient training. Lifting predicates and crash framing make the specification easy to use for developers, and logically atomic crash specifications allow for modular reasoning in GoJournal, making the proof tractable despite complex concurrency and crash interleavings. Submitted November 12, 2021 Accepted January 20, 2022. Manuela will present examples and discuss the scope of AI in her research in the finance domain. While several new GNN architectures have been proposed, the scale of real-world graphsin many cases billions of nodes and edgesposes challenges during model training. SanRazor adopts a novel hybrid approach it captures both dynamic code coverage and static data dependencies of checks, and uses the extracted information to perform a redundant check analysis. We also verified a simple NFS server using GoJournals specs, which confirms that they are helpful for application verification: a significant part of the proof doesnt have to consider concurrency and crashes. We present NrOS, a new OS kernel with a safer approach to synchronization that runs many POSIX programs. We focus on NVMe storage devices and show that it is natural to express these semantics in the kernel and the application and only requires a modest two-bit change to the device interface. DistAI: Data-Driven Automated Invariant Learning for Distributed Protocols Jianan Yao, Runzhou Tao, Ronghui Gu, Jason Nieh . Thanks to selective profiling, DMons profiling overhead is 1.36% on average, making it feasible for production use. For more details on the submission process, and for templates to use with LaTeX, Word, etc., authors should consult the detailed submission requirements. 23 artifacts received the Artifacts Functional badge (88%). Consensus bugs are extremely rare but can be exploited for network split and theft, which cause reliability and security-critical issues in the Ethereum ecosystem. In 2023 I started another two-year term on the . Sanitizers detect unsafe actions such as invalid memory accesses by inserting checks that are validated during a programs execution. As increasingly more sensitive data is being collected to gain valuable insights, the need to natively integrate privacy controls in data analytics frameworks is growing in importance. Pollux improves scheduling performance in deep learning (DL) clusters by adaptively co-optimizing inter-dependent factors both at the per-job level and at the cluster-wide level. Session Chairs: Dushyanth Narayanan, Microsoft Research, and Gala Yadgar, TechnionIsrael Institute of Technology, Jinhyung Koo, Junsu Im, Jooyoung Song, and Juhyung Park, DGIST; Eunji Lee, Soongsil University; Bryan S. Kim, Syracuse University; Sungjin Lee, DGIST. We prove that DistAI is guaranteed to find the -free inductive invariant that proves the desired safety properties in finite time, if one exists. See the Preview Session page for an overview of the topics covered in the program. And yet, they continue to rely on centralized search engines and indexers to help users access the content they seek and navigate the apps. Her robot soccer teams have been RoboCup world champions several times, and the CoBot mobile robots have autonomously navigated for more than 1,000km in university buildings. In the Ethereum network, decentralized Ethereum clients reach consensus through transitioning to the same blockchain states according to the Ethereum specification. All submissions will be treated as confidential prior to publication on the USENIX OSDI 21 website; rejected submissions will be permanently treated as confidential. Second, it innovates on the underlying cryptographic machinery and constructs a new private information retrieval scheme, FastPIR, that reduces the time to process oblivious access requests for mailboxes. To evaluate the security guarantees of Storm, we build a formally verified reference implementation using the Labeled IO (LIO) IFC framework. Existing systems that hide voice call metadata either require trusted intermediaries in the network or scale to only tens of users. After request completion, an I/O device must decide either to minimize latency by immediately firing an interrupt or to optimize for throughput by delaying the interrupt, anticipating that more requests will complete soon and help amortize the interrupt cost. The 20th ACM Workshop on Hot Topics in Networks (HotNets 2021) will bring together researchers in computer networks and systems to engage in a lively debate on the theory and practice of computer networking. Academic and industrial participants present research and experience papers that cover the full range of theory and practice of computer . They collectively make the backup fresh, columnar, and fault-tolerant, even facing millions of concurrent transactions per second. Camera-ready submission (all accepted papers): 2 April 2021; Main conference program: 27-28 April 2021; All deadline times are . Mothy received a PhD in 1995 from the Computer Laboratory of the University of Cambridge, where he was a principal designer and builder of the Nemesis OS. This paper presents the design and implementation of CLP, a tool capable of losslessly compressing unstructured text logs while enabling fast searches directly on the compressed data. Instead of choosing among a small number of known algorithms, our approach searches in a "policy space" of fine-grained actions, resulting in novel algorithms that can outperform existing algorithms by specializing to a given workload. In experiments with real DL jobs and with trace-driven simulations, Pollux reduces average job completion times by 37-50% relative to state-of-the-art DL schedulers, even when they are provided with ideal resource and training configurations for every job. We present selective profiling, a technique that locates data locality problems with low-enough overhead that is suitable for production use. There are two major GNN training obstacles: 1) it relies on high-end servers with many GPUs which are expensive to purchase and maintain, and 2) limited memory on GPUs cannot scale to today's billion-edge graphs. Based on the observation that real-world workloads always feature skewed access patterns, Nap introduces a NUMA-aware layer (NAL) on the top of existing concurrent PM indexes, and steers accesses to hot items to this layer. We observe that scalability challenges in training GNNs are fundamentally different from that in training classical deep neural networks and distributed graph processing; and that commonly used techniques, such as intelligent partitioning of the graph do not yield desired results. HotCRP.com signin Sign in using your HotCRP.com account. Authors should email the program co-chairs, osdi21chairs@usenix.org, a copy of the related workshop paper and a short explanation of the new material in the conference paper beyond that published in the workshop version. Submissions violating the detailed formatting and anonymization rules will not be considered for review. Paper Submission Information All submissions must be received by 11:59 PM AoE (UTC-12) on the day of the corresponding deadline. To help more profitably utilize sanitizers, we introduce SanRazor, a practical tool aiming to effectively detect and remove redundant sanitizer checks. Upon these two primitives, our system can scale to thousands of concurrent enclaves with high resource utilization and eliminate the high-cost initialization of secure memory using fork-style enclave creation without weakening the security guarantees. Zeph enforces privacy policies cryptographically and ensures that data available to third-party applications complies with users' privacy policies. The conference papers and full proceedings are available to registered attendees now and will be available to everyone beginning Wednesday, July 14, 2021. It then feeds those invariants and the desired safety properties to an SMT solver to check if the conjunction of the invariants and the safety properties is inductive. However, the existing one-size-fits-all GNN implementations are insufficient to catch up with the evolving GNN architectures, the ever-increasing graph size, and the diverse node embedding dimensionality. Jiang Zhang, University of Southern California; Shuai Wang, HKUST; Manuel Rigger, Pinjia He, and Zhendong Su, ETH Zurich. Extensive experiments show that GNNAdvisor outperforms the state-of-the-art GNN computing frameworks, such as Deep Graph Library (3.02 faster on average) and NeuGraph (up to 4.10 faster), on mainstream GNN architectures across various datasets. To achieve low overhead, selective profiling gathers runtime execution information selectively and incrementally. Attaching supplementary material is optional; if your paper says that you have source code or formal proofs, you need not attach them to convince the PC of their existence. Horcruxs JavaScript scheduler then uses this information to judiciously parallelize JavaScript execution on the client-side so that the end-state is identical to that of a serial execution, while minimizing coordination and offloading overheads. Submitted papers must be no longer than 12 single-spaced 8.5 x 11 pages, including figures and tables, plus as many pages as needed for references, using 10-point type on 12-point (single-spaced) leading, two-column format, Times Roman or a similar font, within a text block 7 wide x 9 deep. Metadata from voice calls, such as the knowledge of who is communicating with whom, contains rich information about peoples lives. If your accepted paper should not be published prior to the event, please notify production@usenix.org. By monitoring the status of each job during training, Pollux models how their goodput (a novel metric we introduce that combines system throughput with statistical efficiency) would change by adding or removing resources. Pollux: Co-adaptive Cluster Scheduling for Goodput-Optimized Deep Learning, Oort: Efficient Federated Learning via Guided Participant Selection, PET: Optimizing Tensor Programs with Partially Equivalent Transformations and Automated Corrections, Modernizing File System through In-Storage Indexing, Nap: A Black-Box Approach to NUMA-Aware Persistent Memory Indexes, Rearchitecting Linux Storage Stack for s Latency and High Throughput, Optimizing Storage Performance with Calibrated Interrupts, ZNS+: Advanced Zoned Namespace Interface for Supporting In-Storage Zone Compaction, DMon: Efficient Detection and Correction of Data Locality Problems Using Selective Profiling, CLP: Efficient and Scalable Search on Compressed Text Logs, Polyjuice: High-Performance Transactions via Learned Concurrency Control, Retrofitting High Availability Mechanism to Tame Hybrid Transaction/Analytical Processing, The nanoPU: A Nanosecond Network Stack for Datacenters, Beyond malloc efficiency to fleet efficiency: a hugepage-aware memory allocator, Scalable Memory Protection in the PENGLAI Enclave, NrOS: Effective Replication and Sharing in an Operating System, Addra: Metadata-private voice communication over fully untrusted infrastructure, Bringing Decentralized Search to Decentralized Services, Finding Consensus Bugs in Ethereum via Multi-transaction Differential Fuzzing, MAGE: Nearly Zero-Cost Virtual Memory for Secure Computation, Zeph: Cryptographic Enforcement of End-to-End Data Privacy, It's Time for Operating Systems to Rediscover Hardware, DistAI: Data-Driven Automated Invariant Learning for Distributed Protocols, GoJournal: a verified, concurrent, crash-safe journaling system, STORM: Refinement Types for Secure Web Applications, Horcrux: Automatic JavaScript Parallelism for Resource-Efficient Web Computation, SANRAZOR: Reducing Redundant Sanitizer Checks in C/C++ Programs, Dorylus: Affordable, Scalable, and Accurate GNN Training with Distributed CPU Servers and Serverless Threads, GNNAdvisor: An Adaptive and Efficient Runtime System for GNN Acceleration on GPUs, Marius: Learning Massive Graph Embeddings on a Single Machine, P3: Distributed Deep Graph Learning at Scale. Performance experiments show that GoNFS provides similar performance (e.g., at least 90% throughput across several benchmarks on an NVMe disk) to Linuxs NFS server exporting an ext4 file system, suggesting that GoJournal is a competitive journaling system. This kernel is scaled across NUMA nodes using node replication, a scheme inspired by state machine replication in distributed systems. This paper demonstrates that it is possible to achieve s-scale latency using Linux kernel storage stack, even when tens of latency-sensitive applications compete for host resources with throughput-bound applications that perform read/write operations at throughput close to hardware capacity. For general conference information, see https://www.usenix.org/conference/osdi22. Marius is open-sourced at www.marius-project.org. Second, GNNAdvisor implements a novel and highly-efficient 2D workload management tailored for GNN computation to improve GPU utilization and performance under different application settings. Advisor: You have a past or present association as thesis advisor or advisee. The key insight in blk-switch is that Linux's multi-queue storage design, along with multi-queue network and storage hardware, makes the storage stack conceptually similar to a network switch. Dorylus is up to 3.8 faster and 10.7 cheaper compared to existing sampling-based systems. Her specialties include network routing protocols and network security. If your paper is accepted and you need an invitation letter to apply for a visa to attend the conference, please contact conference@usenix.org as soon as possible. Table of Contents | We present Nap, a black-box approach that converts concurrent persistent memory (PM) indexes into NUMA-aware counterparts. PLDI seeks outstanding research that extends and/or applies programming-language concepts to advance the field of computing. Finding the inductive invariant of the distributed protocol is a critical step in verifying the correctness of distributed systems, but takes a long time to do even for simple protocols. She also invented the spanning tree algorithm, which transformed Ethernet from a technology that supported a few hundred nodes, to something that can support large networks. To this end, we propose GNNAdvisor, an adaptive and efficient runtime system to accelerate various GNN workloads on GPU platforms. OSDI'20: 14th USENIX Conference on Operating Systems Design and ImplementationNovember 4 - 6, 2020 ISBN: 978-1-939133-19-9 Published: 04 November 2020 Sponsors: ORACLE, VMware, Google Inc., Amazon, Microsoft Get Alerts for this Conference Save to Binder Export Citation Bibliometrics Citation count 96 Downloads (6 weeks) 317 Downloads (12 months) We present Storm, a web framework that allows developers to build MVC applications with compile-time enforcement of centrally specified data-dependent security policies. Many application domains can benefit from hybrid transaction/analytical processing (HTAP) by executing queries on real-time datasets produced by concurrent transactions. DistAI generates data by simulating the distributed protocol at different instance sizes and recording states as samples. Unfortunately, because devices lack the semantic information about which I/O requests are latency-sensitive, these heuristics can sometimes lead to disastrous results.