Search

Scheduling virtual machines (VMs) to hosts in cloud data centers dictates efficiency and is an NP-hard problem with incomplete information. Prior work improved VM scheduling with predicted VM lifetimes. Our work further improves lifetime-aware scheduling using repredictions with lifetime distributions vs. one-shot prediction. The approach repredicts and adjusts VM and host lifetimes when incorrect predictions emerge. We also present novel approaches for defragmentation and regular system maintenance, which are essential to our data center reliability and optimizations, and are unexplored in prior work. We show that repredictions deliver a fundamental advance in effectiveness over one-shot prediction. We call our novel combination of distribution-based lifetime predictions and scheduling algorithms Lifetime Aware VM Allocation (LAVA). LAVA improves resource stranding and the number of empty hosts, which are critical for large VM scheduling, cloud system updates, and reducing dynamic energy consumption. Our approach runs in production within Google's hyperscale cloud data centers, where it improves efficiency by decreasing stranded compute and memory resources by ~3% and ~2% respectively, and increases availability for large VMs and cloud system updates by increasing empty hosts by 2.3-9.2 pp in production. We also show a reduction in VM migrations for host defragmentation and maintenance. In addition to our fleet-wide production deployment, we perform simulation studies to characterize the design space and show that our algorithm significantly outperforms the state of the art lifetime-based scheduling approach.

Production garbage collectors make substantial compromises in pursuit of reduced pause times. They require far more CPU cycles and memory than prior simpler collectors. concurrent copying collectors (C4, ZGC, and Shenandoah) suffer from the following design limitations. 1) Concurrent copying. They only reclaim memory by copying, which is inherently expensive with high memory bandwidth demands. Concurrent copying also requires expensive read and write barriers. 2) Scalability. They depend on tracing, which in the limit and in practice does not scale. 3) Immediacy. They do not reclaim older objects promptly, incurring high memory overheads. We present LXR, which takes a very different approach to optimizing responsiveness and throughput by minimizing concurrent collection work and overheads. 1) LXR reclaims most memory without any copying by using the Immix heap structure. It then combats fragmentation with limited judicious stop-the-world copying. 2) LXR uses reference counting to achieve both scalability and immediacy, promptly reclaiming young and old objects. It uses concurrent tracing as needed for identifying cyclic garbage. 3) To minimize pause times while allowing judicious copying of mature objects, LXR introduces remembered sets for reference counting and concurrent decrement processing. 4) LXR introduces a novel low-overhead write barrier that combines coalescing reference counting, concurrent tracing, and remembered set maintenance. The result is a collector with excellent responsiveness and throughput. On the widely-used Lucene search engine with a generously sized heap, LXR has 6x higher throughput while delivering 30x lower 99.9 percentile tail latency than the popular Shenandoah production collector in its default configuration., Comment: 17 pages, 7 Figures. This extends the original publication with an LBO analysis (Section 5.5)

Modern science, technology, and politics are all permeated by data that comes from people, measurements, or computational processes. While this data is often incomplete, corrupt, or lacking in sufficient accuracy and precision, explicit consideration of uncertainty is rarely part of the computational and decision making pipeline. The CCC Workshop on Quantification, Communication, and Interpretation of Uncertainty in Simulation and Data Science explored this problem, identifying significant shortcomings in the ways we currently process, present, and interpret uncertain data. Specific recommendations on a research agenda for the future were made in four areas: uncertainty quantification in large-scale computational simulations, uncertainty quantification in data science, software support for uncertainty computation, and better integration of uncertainty quantification and communication to stakeholders., Comment: A Computing Community Consortium (CCC) workshop report, 28 pages

Although the computer science community successfully harnessed exponential increases in computer performance to drive societal and economic change, the exponential growth in publications is proving harder to accommodate. To gain a deeper understanding of publication growth and inform how the computer science community should handle this growth, we analyzed publication practices from several perspectives: ACM sponsored publications in the ACM Digital Library as a whole: subdisciplines captured by ACM's Special Interest Groups (SIGs); ten top conferences; institutions; four top U.S. departments; authors; faculty; and PhDs between 1990 and 2012. ACM publishes a large fraction of all computer science research. We first summarize how we believe our main findings inform (1) expectations on publication growth, (2) how to distinguish research quality from output quantity; and (3) the evaluation of individual researchers. We then further motivate the study of computer science publication practices and describe our methodology and results in detail.

Stream analytics have an insatiable demand for memory and performance. Emerging hybrid memories combine commodity DDR4 DRAM with 3D-stacked High Bandwidth Memory (HBM) DRAM to meet such demands. However, achieving this promise is challenging because (1) HBM is capacity-limited and (2) HBM boosts performance best for sequential access and high parallelism workloads. At first glance, stream analytics appear a particularly poor match for HBM because they have high capacity demands and data grouping operations, their most demanding computations, use random access. This paper presents the design and implementation of StreamBox-HBM, a stream analytics engine that exploits hybrid memories to achieve scalable high performance. StreamBox-HBM performs data grouping with sequential access sorting algorithms in HBM, in contrast to random access hashing algorithms commonly used in DRAM. StreamBox-HBM solely uses HBM to store Key Pointer Array (KPA) data structures that contain only partial records (keys and pointers to full records) for grouping operations. It dynamically creates and manages prodigious data and pipeline parallelism, choosing when to allocate KPAs in HBM. It dynamically optimizes for both the high bandwidth and limited capacity of HBM, and the limited bandwidth and high capacity of standard DRAM. StreamBox-HBM achieves 110 million records per second and 238 GB/s memory bandwidth while effectively utilizing all 64 cores of Intel's Knights Landing, a commercial server with hybrid memory. It outperforms stream engines with sequential access algorithms without KPAs by 7x and stream engines with random access algorithms by an order of magnitude in throughput. To the best of our knowledge, StreamBox-HBM is the first stream engine optimized for hybrid memories.

Non-volatile memory (NVM) has the potential to disrupt the boundary between memory and storage, including the abstractions that manage this boundary. Researchers comparing the speed, durability, and abstractions of hybrid systems with DRAM, NVM, and disk to traditional systems typically use simulation, which makes it easy to evaluate different hardware technologies and parameters. Unfortunately, simulation is extremely slow, limiting the number of applications and dataset sizes in the evaluation. Simulation typically precludes realistic multiprogram workloads and considering runtime and operating system design alternatives. Good methodology embraces a variety of techniques for validation, expanding the experimental scope, and uncovering new insights. This paper introduces an emulation platform for hybrid memory that uses commodity NUMA servers. Emulation complements simulation well, offering speed and accuracy for realistic workloads, and richer software experimentation. We use a thread-local socket to emulate DRAM and the remote socket to emulate NVM. We use standard C library routines to allocate heap memory in the DRAM or NVM socket for use with explicit memory management or garbage collection. We evaluate the emulator using various configurations of write-rationing garbage collectors that improve NVM lifetimes by limiting writes to NVM, and use 15 applications from three benchmark suites with various datasets and workload configurations. We show emulation enhances simulation results. The two systems confirm most trends, such as NVM write and read rates of different software configurations, increasing our confidence for predicting future system effects. Emulation adds novel insights, such as the non-linear effects of multi-program workloads on write rates.

Rankings are a fact of life. Whether or not one likes them, they exist and are influential. Within academia, and in computer science in particular, rankings not only capture our attention but also widely influence people who have a limited understanding of computing science research, including prospective students, university administrators, and policy-makers. In short, rankings matter. This position paper advocates for the adoption of "GOTO rankings": rankings that use Good data, are Open, Transparent, and Objective, and the rejection of rankings that do not meet these criteria., Comment: Accepted, to appear in Communications of the ACM

Sensors on mobile devices---accelerometers, gyroscopes, pressure meters, and GPS---invite new applications in gesture recognition, gaming, and fitness tracking. However, programming them remains challenging because human gestures captured by sensors are noisy. This paper illustrates that noisy gestures degrade training and classification accuracy for gesture recognition in state-of-the-art deterministic Hidden Markov Models (HMM). We introduce a new statistical quantization approach that mitigates these problems by (1) during training, producing gesture-specific codebooks, HMMs, and error models for gesture sequences; and (2) during classification, exploiting the error model to explore multiple feasible HMM state sequences. We implement classification in Uncertain, a probabilistic programming system that encapsulates HMMs and error models and then automates sampling and inference in the runtime. Uncertain developers directly express a choice of application-specific trade-off between recall and precision at gesture recognition time, rather than at training time. We demonstrate benefits in configurability, precision, recall, and recognition on two data sets with 25 gestures from 28 people and 4200 total gestures. Incorporating gesture error more accurately in modeling improves the average recognition rate of 20 gestures from 34\% in prior work to 62\%. Incorporating the error model during classification further improves the average gesture recognition rate to 71\%. As far as we are aware, no prior work shows how to generate an HMM error model during training and use it to improve classification rates.

Discusses various concepts of crisis intervention theory as applied to the treatment of rural victims of child sexual abuse and their families. Describes a working example of a rural child sexual abuse crisis group, including the initial stages of group development, the cohesive/working stage, and the termination stage. (LP)