Universidade Federal da Bahia（巴伊亚联邦大学）；Universitat Politècnica de Catalunya（加泰罗尼亚理工大学）；Barcelona Supercomputing Center（巴塞罗那超级计算中心）；University of Pittsburgh（匹兹堡大学）；Norwegian University of Science and Technology（挪威科技大学）
Felippe V. Zacarias, Vinicius Petrucci, Rajiv Nishtala, Paul Carpenter, Daniel Mossé
备注：
Submitted to Journal of Parallel and Distributed Computing
Many HPC applications suffer from a bottleneck in the shared caches, instruction execution units, I/O or memory bandwidth, even though the remaining resources may be underutilized. It is hard for developers and runtime systems to ensure that all critical resources are fully exploited by a single application, so an attractive technique for increasing HPC system utilization is to colocate multiple applications on the same server. When applications share critical resources, however, contention on shared resources may lead to reduced application performance.
In this paper, we show that server efficiency can be improved by first modeling the expected performance degradation of colocated applications based on measured hardware performance counters, and then exploiting the model to determine an optimized mix of colocated applications. This paper presents a new intelligent resource manager and makes the following contributions: (1) a new machine learning model to predict the performance degradation of colocated applications based on hardware counters and (2) an intelligent scheduling scheme deployed on an existing resource manager to enable application co-scheduling with minimum performance degradation. Our results show that our approach achieves performance improvements of 7% (avg) and 12% (max) compared to the standard policy commonly used by existing job managers.
In hosting environments such as IaaS clouds, desirable application performance is usually guaranteed through the use of Service Level Agreements (SLAs), which specify minimal fractions of resource capacities that must be allocated for use for proper operation. Arbitrary colocation of applications with different SLAs on a single host may result in inefficient utilization of the host's resources. In this paper, we propose that periodic resource allocation and consumption models be used for a more granular expression of SLAs. Our proposed SLA model has the salient feature that it exposes flexibil…
Datacenters suffer from resource utilization inefficiencies due to the conflicting goals of service owners and platform providers. Service owners intending to maintain Service Level Objectives (SLO) for themselves typically request a conservative amount of resources. Platform providers want to increase operational efficiency to reduce capital and operating costs. Achieving both operational efficiency and SLO for individual services at the same time is challenging due to the diversity in service workload characteristics, resource usage patterns that are dependent on input load, heterogeneity in…
We present AccaSim, a simulator for workload management in HPC systems. Thanks to AccaSim's scalability to large workload datasets, support for easy customization, and practical automated tools to aid experimentation, users can easily represent various real HPC systems, develop novel advanced dispatchers and evaluate them in a convenient way across different workload sources. AccaSim is thus an attractive tool for conducting job dispatching research in HPC systems.
Traditionally, on-demand, rigid, and malleable applications have been scheduled and executed on separate systems. The ever-growing workload demands and rapidly developing HPC infrastructure trigger the interest of converging these applications on a single HPC system. Although allocating the hybrid workloads within one system could potentially improve system efficiency, it is difficult to balance the tradeoff between the responsiveness of on-demand requests, the incentive for malleable jobs, and the performance of rigid applications. In this study, we present several scheduling mechanisms to ad…
Workload characterization is an integral part of performance analysis of high performance computing (HPC) systems. An understanding of workload properties sheds light on resource utilization and can be used to inform performance optimization both at the software and system configuration levels. It can provide information on how computational science usage modalities are changing that could potentially aid holistic capacity planning for the wider HPC ecosystem. Here, we report on the results of a detailed workload analysis of the portfolio of supercomputers comprising the NSF Innovative HPC pro…
High performance computing (HPC) applications have a high requirement on storage speed and capacity. Non-volatile memory is a promising technology to replace traditional storage devices to improve HPC performance. Earlier in 2017, Intel and Micron released first NVM product -- Intel Optane SSDs. Optane is much faster and more durable than the traditional storage device. It creates a bridge to narrow the performance gap between DRAM and storage. But is the existing HPC I/O stack still suitable for new NVM devices like Intel Optane? How does HPC I/O workload perform with Intel Optane? In this pa…
Containerization technology offers lightweight OS-level virtualization, and enables portability, reproducibility, and flexibility by packing applications with low performance overhead and low effort to maintain and scale them. Moreover, container orchestrators (e.g., Kubernetes) are widely used in the Cloud to manage large clusters running many containerized applications. However, scheduling policies that consider the performance nuances of containerized High Performance Computing (HPC) workloads have not been well-explored yet. This paper conducts fine-grained scheduling policies for containe…
Illinois Institute of Technology（伊利诺伊理工学院）；Argonne National Laboratory（阿贡国家实验室）；Northern Illinois University（北伊利诺伊大学）
Cluster scheduler is crucial in high-performance computing (HPC). It determines when and which user jobs should be allocated to available system resources. Existing cluster scheduling heuristics are developed by human experts based on their experience with specific HPC systems and workloads. However, the increasing complexity of computing systems and the highly dynamic nature of application workloads have placed tremendous burden on manually designed and tuned scheduling heuristics. More aggressive optimization and automation are needed for cluster scheduling in HPC. In this work, we present a…
There is increasing interest in the use of HPC machines for urgent workloads to help tackle disasters as they unfold. Whilst batch queue systems are not ideal in supporting such workloads, many disadvantages can be worked around by accurately predicting when a waiting job will start to run. However there are numerous challenges in achieving such a prediction with high accuracy, not least because the queue's state can change rapidly and depend upon many factors. In this work we explore a novel machine learning approach for predicting queue wait times, hypothesising that such a model can capture…
The primary motivation for uptake of virtualization has been resource isolation, capacity management and resource customization allowing resource providers to consolidate their resources in virtual machines. Various approaches have been taken to integrate virtualization in to scientific Grids especially in the arena of High Performance Computing (HPC) to run grid jobs in virtual machines, thus enabling better provisioning of the underlying resources and customization of the execution environment on runtime. Despite the gains, virtualization layer also incur a performance penalty and its not ve…
Energy consumption is increasingly becoming a limiting factor to the design of faster large-scale parallel systems, and development of energy-efficient and energy-aware applications is today a relevant issue for HPC code-developer communities. In this work we focus on energy performance of the Knights Landing (KNL) Xeon Phi, the latest many-core architecture processor introduced by Intel into the HPC market. We take into account the 64-core Xeon Phi 7230, and analyze its energy performance using both the on-chip MCDRAM and the regular DDR4 system memory as main storage for the application data…
Containerisation demonstrates its efficiency in application deployment in cloud computing. Containers can encapsulate complex programs with their dependencies in isolated environments, hence are being adopted in HPC clusters. HPC workload managers lack micro-services support and deeply integrated container management, as opposed to container orchestrators (e.g. Kubernetes). We introduce Torque-Operator (a plugin) which serves as a bridge between HPC workload managers and container Orchestrators.
In this paper, we analyze the performance and energy consumption of an Arm-based high-performance computing (HPC) system developed within the European project Mont-Blanc 3. This system, called Dibona, has been integrated by ATOS/Bull, and it is powered by the latest Marvell's CPU, ThunderX2. This CPU is the same one that powers the Astra supercomputer, the first Arm-based supercomputer entering the Top500 in November 2018. We study from micro-benchmarks up to large production codes. We include an interdisciplinary evaluation of three scientific applications (a finite-element fluid dynamics cod…
Social influence among users (e.g., collaboration on a project) creates bursty behavior in the underlying high performance computing (HPC) workloads. Using representative HPC and cluster workload logs, this paper identifies, analyzes, and quantifies the level of social influence across HPC users. We show the existence of a social graph that is characterized by a pattern of dominant users and followers. This pattern also follows a power-law distribution, which is consistent with those observed in mainstream social networks. Given its potential impact on HPC workloads prediction and scheduling, …
New HPC machines are getting close to the exascale. Power consumption for those machines has been increasing, and researchers are studying ways to reduce it. A second trend is HPC machines' growing complexity, with increasing heterogeneous hardware components and different clusters architectures cooperating in the same machine. We refer to these environments with the term heterogeneous multi-cluster environments. With the aim of optimizing performance and energy consumption in these environments, this paper proposes an Energy-Aware-Multi-Cluster (EAMC) job scheduling policy. EAMC-policy is abl…
Tools-CC-Bench: a Benchmark Suite for Collective Communication with Compression in HPC and AI Workloads
Institute of Computing Technology, Chinese Academy of Sciences（中国科学院计算技术研究所）；College of Computer Science and Technology, National University of Defense Technology（国防科技大学计算机学院）；School of Computer Science, Nanjing University（南京大学计算机系）
The HPC community is actively researching and evaluating tools to support execution of scientific applications in cloud-based environments. Among the various technologies, containers have recently gained importance as they have significantly better performance compared to full-scale virtualization, support for microservices and DevOps, and work seamlessly with workflow and orchestration tools. Docker is currently the leader in containerization technology because it offers low overhead, flexibility, portability of applications, and reproducibility. Singularity is another container solution that…
Modern HPC workload managers and their careful tuning contribute to the high utilization of HPC clusters. However, due to inevitable uncertainty it is impossible to completely avoid node idleness. Although such idle slots are usually too short for any HPC job, they are too long to ignore them. Function-as-a-Service (FaaS) paradigm promisingly fills this gap, and can be a good match, as typical FaaS functions last seconds, not hours. Here we show how to build a FaaS infrastructure on idle nodes in an HPC cluster in such a way that it does not affect the performance of the HPC jobs significantly…
The use of cloud computational resources has become increasingly important for companies and researchers to access on-demand and at any moment high-performance resources. However, given the wide variety of virtual machine types, network configurations, number of instances, among others, finding the best configuration that reduces costs and resource waste while achieving acceptable performance is a hard task even for specialists. Thus, many approaches to find these good or optimal configurations for a given program have been proposed. Observing the performance of an application in some configur…
Object storage solutions potentially address long-standing performance issues with POSIX file systems for certain I/O workloads, and new storage technologies offer promising performance characteristics for data-intensive use cases. In this work, we present a preliminary assessment of Intel's Distributed Asynchronous Object Store (DAOS), an emerging high-performance object store, in conjunction with non-volatile storage and evaluate its potential use for HPC storage. We demonstrate DAOS can provide the required performance, with bandwidth scaling linearly with additional DAOS server nodes in mo…
Cognitive simulation (CogSim) is an important and emerging workflow for HPC scientific exploration and scientific machine learning (SciML). One challenging workload for CogSim is the replacement of one component in a complex physical simulation with a fast, learned, surrogate model that is "inside" of the computational loop. The execution of this in-the-loop inference is particularly challenging because it requires frequent inference across multiple possible target models, can be on the simulation's critical path (latency bound), is subject to requests from multiple MPI ranks, and typically co…
Software container solutions have revolutionized application development approaches by enabling lightweight platform abstractions within the so-called "containers." Several solutions are being actively developed in attempts to bring the benefits of containers to high-performance computing systems with their stringent security demands on the one hand and fundamental resource sharing requirements on the other. In this paper, we discuss the benefits and short-comings of such solutions when deployed on real HPC systems and applied to production scientific applications.We highlight use cases that a…
Cloud computing represents an appealing opportunity for cost-effective deployment of HPC workloads on the best-fitting hardware. However, although cloud and on-premise HPC systems offer similar computational resources, their network architecture and performance may differ significantly. For example, these systems use fundamentally different network transport and routing protocols, which may introduce network noise that can eventually limit the application scaling. This work analyzes network performance, scalability, and cost of running HPC workloads on cloud systems. First, we consider latency…
Resource demands of HPC applications vary significantly. However, it is common for HPC systems to primarily assign resources on a per-node basis to prevent interference from co-located workloads. This gap between the coarse-grained resource allocation and the varying resource demands can lead to HPC resources being not fully utilized. In this study, we analyze the resource usage and application behavior of NERSC's Perlmutter, a state-of-the-art open-science HPC system with both CPU-only and GPU-accelerated nodes. Our one-month usage analysis reveals that CPUs are commonly not fully utilized, e…
Engineering is an important domain for supercomputing, with the Alya model being a popular code for undertaking such simulations. With ever increasing demand from users to model larger, more complex systems at reduced time to solution it is important to explore the role that novel hardware technologies, such as FPGAs, can play in accelerating these workloads on future exascale systems. In this paper we explore the porting of Alya's incompressible flow matrix assembly kernel, which accounts for a large proportion of the model runtime, onto FPGAs. After describing in detail successful strategies…
Data stores are the foundation on which data science, in all its variations, is built upon. They provide a queryable interface to structured and unstructured data. Data science often starts by leveraging these query features to perform initial data preparation. However, most data stores are designed to run continuously to service disparate user requests with little or no downtime. Many HPC architectures process user requests by job queue scheduler and maintain a shard filesystem to store a jobs persistent data. We deploy a MongoDB sharded cluster with a run script that is designed to run a dat…
