DOCSLIB.ORG

1 Cluster Computing at a Glance Chapter 1: by M. Baker and R. Buyya Introduction How to Run Applications Faster ? Era of Computi

Total Page：16

File Type：pdf, Size：1020Kb

Download full-text PDF Read full-text
1 Cluster Computing at a Glance Chapter 1: by M. Baker and R. Buyya Introduction How to Run Applications Faster ? Era of Computi Cluster Computing at a Glance Chapter 1: by M. Baker and R. Buyya High Performance Cluster Computing Introduction Scalable Parallel Computer Architecture Architectures and Systems Towards Low Cost Parallel Computing and Motivations Windows of Opportunity A Cluster Computer and its Architecture Clusters Classifications Book Editor: Rajkumar Buyya Commodity Components for Clusters Network Service/Communications SW Slides Prepared by: Hai Jin Cluster Middleware and Single System Image Resource Management and Scheduling (RMS) Programming Environments and Tools Cluster Applications Internet and Cluster Computing Center Representative Cluster Systems Cluster of SMPs (CLUMPS) Summary and Conclusions 2 http://www.buyya.com/cluster/ Introduction How to Run Applications Faster ? Need more computing power There are 3 ways to improve performance: Improve the operating speed of processors & Work Harder other components Work Smarter constrained by the speed of light, thermodynamic laws, & the high financial costs for processor Get Help fabrication Computer Analogy Using faster hardware Connect multiple processors together & coordinate their computational efforts Optimized algorithms and techniques used to solve computational tasks parallel computers allow the sharing of a computational task among Multiple computers to solve a particular task multiple processors 3 4 Era of Computing Two Eras of Computing Rapid technical advances the recent advances in VLSI technology software technology OS, PL, development methodologies, & tools grand challenge applications have become the main driving force Parallel computing one of the best ways to overcome the speed bottleneck of a single processor good price/performance ratio of a small cluster- based parallel computer 5 6 1 Scalable Parallel Computer Scalable Parallel Computer Architectures Architectures Taxonomy MPP A large parallel processing system with a shared-nothing based on how processors, memory & architecture interconnect are laid out Consist of several hundred nodes with a high-speed Massively Parallel Processors (MPP) interconnection network/switch Each node consists of a main memory & one or more Symmetric Multiprocessors (SMP) processors Cache-Coherent Nonuniform Memory Access Runs a separate copy of the OS (CC-NUMA) SMP Distributed Systems 2-64 processors today Shared-everything architecture Clusters All processors share all the global resources available Single copy of the OS runs on these systems 7 8 Scalable Parallel Computer Key Characteristics of Architectures Scalable Parallel Computers CC-NUMA a scalable multiprocessor system having a cache-coherent nonuniform memory access architecture every processor has a global view of all of the memory Distributed systems considered conventional networks of independent computers have multiple system images as each node runs its own OS the individual machines could be combinations of MPPs, SMPs, clusters, & individual computers Clusters a collection of workstations of PCs that are interconnected by a high-speed network work as an integrated collection of resources have a single system image spanning all its nodes 9 10 Towards Low Cost Parallel Motivations of using NOW over Computing Specialized Parallel Computers Parallel processing Individual workstations are becoming increasing linking together 2 or more computers to jointly solve some computational problem powerful since the early 1990s, an increasing trend to move away Communication bandwidth between workstations is from expensive and specialized proprietary parallel supercomputers towards networks of workstations increasing and latency is decreasing the rapid improvement in the availability of commodity high Workstation clusters are easier to integrate into performance components for workstations and networks existing networks Low-cost commodity supercomputing Typical low user utilization of personal workstations from specialized traditional supercomputing platforms to cheaper, general purpose systems consisting of loosely Development tools for workstations are more mature coupled components built up from single or multiprocessor PCs or workstations Workstation clusters are a cheap and readily need to standardization of many of the tools and utilities available used by parallel applications (ex) MPI, HPF Clusters can be easily grown 11 12 2 Trend Windows of Opportunities Parallel Processing Workstations with UNIX for science & Use multiple processors to build MPP/DSM-like systems for industry vs PC-based machines for parallel computing Network RAM administrative work & work processing Use memory associated with each workstation as aggregate DRAM cache A rapid convergence in processor Software RAID performance and kernel-level functionality Redundant array of inexpensive disks of UNIX workstations and PC-based Use the arrays of workstation disks to provide cheap, highly machines available, & scalable file storage Possible to provide parallel I/O support to applications Use arrays of workstation disks to provide cheap, highly available, and scalable file storage Multipath Communication Use multiple networks for parallel data transfer between 13 14 nodes Cluster Computer and its Architecture Cluster Computer Architecture A cluster is a type of parallel or distributed processing system, which consists of a collection of interconnected stand-alone computers cooperatively working together as a single, integrated computing resource A node a single or multiprocessor system with memory, I/O facilities, & OS generally 2 or more computers (nodes) connected together in a single cabinet, or physically separated & connected via a LAN appear as a single system to users and applications provide a cost-effective way to gain features and benefits 15 16 Prominent Components of Beowulf Cluster Computers (I) Multiple High Performance Computers PCs Head Node Workstations Switch SMPs (CLUMPS) Distributed HPC Systems leading to Metacomputing Client Node1 Client Node2 17 18 3 Prominent Components of Prominent Components of Cluster Computers (II) Cluster Computers (III) High Performance Networks/Switches State of the art Operating Systems Ethernet (10Mbps), Fast Ethernet Linux (Beowulf) (100Mbps), Microsoft NT(Illinois HPVM) SUN Solaris (Berkeley NOW) InfiniteBand (1-8 Gbps) IBM AIX (IBM SP2) Gigabit Ethernet (1Gbps) HP UX (Illinois - PANDA) SCI (Dolphin - MPI- 12micro-sec latency) Mach (Microkernel based OS) (CMU) ATM Cluster Operating Systems (Solaris MC, SCO Myrinet (1.2Gbps) Unixware, MOSIX (academic project) Digital Memory Channel OS gluing layers (Berkeley Glunix) FDDI 19 20 Prominent Components of Prominent Components of Cluster Computers (IV) Cluster Computers (V) Network Interface Card Fast Communication Protocols and Services Myrinet has NIC Active Messages (Berkeley) InfiniteBand (HBA) Fast Messages (Illinois) User-level access support U-net (Cornell) XTP (Virginia) 21 22 Prominent Components of Prominent Components of Cluster Computers (VI) Cluster Computers (VII) Cluster Middleware Parallel Programming Environments and Tools Single System Image (SSI) Threads (PCs, SMPs, NOW..) POSIX Threads System Availability (SA) Infrastructure Java Threads Hardware MPI Linux, NT, on many Supercomputers DEC Memory Channel, DSM (Alewife, DASH), SMP Techniques PVM Operating System Kernel/Gluing Layers Software DSMs (Shmem) Solaris MC, Unixware, GLUnix Compilers C/C++/Java Applications and Subsystems Parallel programming with C++ (MIT Press book) Applications (system management and electronic forms) RAD (rapid application development tools) Runtime systems (software DSM, PFS etc.) GUI based tools for PP modeling Resource management and scheduling software (RMS) Debuggers CODINE, LSF, PBS, NQS, etc. Performance Analysis Tools 23 24 Visualization Tools 4 Prominent Components of Key Operational Benefits of Cluster Computers (VIII) Clustering Applications High Performance Sequential Expandability and Scalability Parallel / Distributed (Cluster-aware app.) High Throughput Grand Challenging applications Weather Forecasting High Availability Quantum Chemistry Molecular Biology Modeling Engineering Analysis (CAD/CAM) ………………. PDBs, web servers,data-mining 25 26 Clusters Classification (I) Clusters Classification (II) Application Target Node Ownership High Performance (HP) Clusters Dedicated Clusters Grand Challenging Applications Non-dedicated clusters High Availability (HA) Clusters Adaptive parallel computing Mission Critical applications Communal multiprocessing 27 28 Clusters Classification (III) Clusters Classification (IV) Node Operating System Node Hardware Linux Clusters (e.g., Beowulf) Clusters of PCs (CoPs) Solaris Clusters (e.g., Berkeley NOW) Piles of PCs (PoPs) NT Clusters (e.g., HPVM) Clusters of Workstations (COWs) AIX Clusters (e.g., IBM SP2) SCO/Compaq Clusters (Unixware) Clusters of SMPs (CLUMPs) Digital VMS Clusters HP-UX clusters Microsoft Wolfpack clusters 29 30 5 Clusters Classification (V) Clusters Classification (VI) Levels of Clustering Node Configuration Group Clusters (#nodes: 2-99) Homogeneous Clusters Nodes are connected by SAN like Myrinet Departmental Clusters (#nodes: 10s to 100s) All nodes will have similar architectures and run the same OSs Organizational Clusters (#nodes: many 100s) National Metacomputers (WAN/Internet- Heterogeneous Clusters based) All nodes will have different architectures
Load more

Recommended publications
  • Sprite File System There Are Three Important Aspects of the Sprite ®Le System: the Scale of the System, Location-Transparency, and Distributed State
    Naming, State Management, and User-Level Extensions in the Sprite Distributed File System Copyright 1990 Brent Ballinger Welch CHAPTER 1 Introduction ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ ¡ This dissertation concerns network computing environments. Advances in network and microprocessor technology have caused a shift from stand-alone timesharing systems to networks of powerful personal computers. Operating systems designed for stand-alone timesharing hosts do not adapt easily to a distributed environment. Resources like disk storage, printers, and tape drives are not concentrated at a single point. Instead, they are scattered around the network under the control of different hosts. New operating system mechanisms are needed to handle this sort of distribution so that users and application programs need not worry about the distributed nature of the underlying system. This dissertation explores the approach of centering a distributed computing environment around a shared network ®le system. The ®le system is chosen as a starting point because it is a heavily used service in stand-alone systems, and the read/write para- digm of the ®le system is a familiar one that can be applied to many system resources. The ®le system described in this dissertation provides a distributed name space for sys- tem resources, and it provides remote access facilities so all resources are available throughout the network. Resources accessible via the ®le system include disk storage, other types of peripheral devices, and user-implemented service applications. The result- ing system is one where resources are named and accessed via the shared ®le system, and the underlying distribution of the system among a collection of hosts is not important to users.
    [Show full text]
  • Clustering with Openmosix
    Clustering with openMosix Maurizio Davini (Department of Physics and INFN Pisa) Presented by Enrico Mazzoni (INFN Pisa) Introduction • What is openMosix? – Single-System Image – Preemptive Process Migration – The openMosix File System (MFS) • Application Fields • openMosix vs Beowulf • The people behind openMosix • The openMosix GNU project • Fork of openMosix code 12/06/2003 HTASC 2 The openMosix Project MileStones • Born early 80s on PDP-11/70. One full PDP and disk-less PDP, therefore process migration idea. • First implementation on BSD/pdp as MS.c thesis. • VAX 11/780 implementation (different word size, different memory architecture) • Motorola / VME bus implementation as Ph.D. thesis in 1993 for under contract from IDF (Israeli Defence Forces) • 1994 BSDi version • GNU and Linux since 1997 • Contributed dozens of patches to the standard Linux kernel • Split Mosix / openMosix November 2001 • Mosix standard in Linux 2.5? 12/06/2003 HTASC 3 What is openMOSIX • Linux kernel extension (2.4.20) for clustering • Single System Image - like an SMP, for: – No need to modify applications – Adaptive resource management to dynamic load characteristics (CPU intensive, RAM intensive, I/O etc.) – Linear scalability (unlike SMP) 12/06/2003 HTASC 4 A two tier technology 1. Information gathering and dissemination – Support scalable configurations by probabilistic dissemination algorithms – Same overhead for 16 nodes or 2056 nodes 2. Pre-emptive process migration that can migrate any process, anywhere, anytime - transparently – Supervised by adaptive
    [Show full text]
  • Distributed Virtual Machines: a System Architecture for Network Computing
    Distributed Virtual Machines: A System Architecture for Network Computing Emin Gün Sirer, Robert Grimm, Arthur J. Gregory, Nathan Anderson, Brian N. Bershad {egs,rgrimm,artjg,nra,bershad}@cs.washington.edu http://kimera.cs.washington.edu Dept. of Computer Science & Engineering University of Washington Seattle, WA 98195-2350 Abstract Modern virtual machines, such as Java and Inferno, are emerging as network computing platforms. While these virtual machines provide higher-level abstractions and more sophisticated services than their predecessors from twenty years ago, their architecture has essentially remained unchanged. State of the art virtual machines are still monolithic, that is, they are comprised of closely-coupled service components, which are thus replicated over all computers in an organization. This crude replication of services forms one of the weakest points in today’s networked systems, as it creates widely acknowledged and well-publicized problems of security, manageability and performance. We have designed and implemented a new system architecture for network computing based on distributed virtual machines. In our system, virtual machine services that perform rule checking and code transformation are factored out of clients and are located in enterprise- wide network servers. The services operate by intercepting application code and modifying it on the fly to provide additional service functionality. This architecture reduces client resource demands and the size of the trusted computing base, establishes physical isolation between virtual machine services and creates a single point of administration. We demonstrate that such a distributed virtual machine architecture can provide substantially better integrity and manageability than a monolithic architecture, scales well with increasing numbers of clients, and does not entail high overhead.
    [Show full text]
  • Workstation Operating Systems Mac OS 9
    15-410 “Now that we've covered the 1970's...” Plan 9 Nov. 25, 2019 Dave Eckhardt 1 L11_P9 15-412, F'19 Overview “The land that time forgot” What style of computing? The death of timesharing The “Unix workstation problem” Design principles Name spaces File servers The TCP file system... Runtime environment 3 15-412, F'19 The Land That Time Forgot The “multi-core revolution” already happened once 1982: VAX-11/782 (dual-core) 1984: Sequent Balance 8000 (12 x NS32032) 1985: Encore MultiMax (20 x NS32032) 1990: Omron Luna88k workstation (4 x Motorola 88100) 1991: KSR1 (1088 x KSR1) 1991: “MCS” paper on multi-processor locking algorithms 1995: BeBox workstation (2 x PowerPC 603) The Land That Time Forgot The “multi-core revolution” already happened once 1982: VAX-11/782 (dual-core) 1984: Sequent Balance 8000 (12 x NS32032) 1985: Encore MultiMax (20 x NS32032) 1990: Omron Luna88k workstation (4 x Motorola 88100) 1991: KSR1 (1088 x KSR1) 1991: “MCS” paper on multi-processor locking algorithms 1995: BeBox workstation (2 x PowerPC 603) Wow! Why was 1995-2004 ruled by single-core machines? What operating systems did those multi-core machines run? The Land That Time Forgot Why was 1995-2004 ruled by single-core machines? In 1995 Intel + Microsoft made it feasible to buy a fast processor that fit on one chip, a fast I/O bus, multiple megabytes of RAM, and an OS with memory protection. Everybody could afford a “workstation”, so everybody bought one. Massive economies of scale existed in the single- processor “Wintel” universe.
    [Show full text]
  • Distributed Operating Systems
    Distributed Operating Systems ANDREW S. TANENBAUM and ROBBERT VAN RENESSE Department of Mathematics and Computer Science, Vrije Universiteit, Amsterdam, The Netherlands Distributed operating systems have many aspects in common with centralized ones, but they also differ in certain ways. This paper is intended as an introduction to distributed operating systems, and especially to current university research about them. After a discussion of what constitutes a distributed operating system and how it is distinguished from a computer network, various key design issues are discussed. Then several examples of current research projects are examined in some detail, namely, the Cambridge Distributed Computing System, Amoeba, V, and Eden. Categories and Subject Descriptors: C.2.4 [Computer-Communications Networks]: Distributed Systems-network operating system; D.4.3 [Operating Systems]: File Systems Management-distributed file systems; D.4.5 [Operating Systems]: Reliability-fault tolerance; D.4.6 [Operating Systems]: Security and Protection-access controls; D.4.7 [Operating Systems]: Organization and Design-distributed systems General Terms: Algorithms, Design, Experimentation, Reliability, Security Additional Key Words and Phrases: File server INTRODUCTION more convenient to use than the bare ma- chine. Examples of well-known centralized Everyone agrees that distributed systems (i.e., not distributed) operating systems are are going to be very important in the future. CP/M,’ MS-DOS,’ and UNIX.3 Unfortunately, not everyone agrees on A distributed operating system is one that what they mean by the term “distributed looks to its users like an ordinary central- system.” In this paper we present a view- ized operating system but runs on multi- point widely held within academia about ple, independent central processing units what is and is not a distributed system, we (CPUs).
    [Show full text]
  • A Single System Image Java Operating System for Sensor Networks
    A SINGLE SYSTEM IMAGE JAVA OPERATING SYSTEM FOR SENSOR NETWORKS Emin Gun Sirer Rimon Barr John C. Bicket Daniel S. Dantas Computer Science Department Cornell University Ithaca, NY 14853 {egs, barr, bicket, ddantas}@cs.cornell.edu Abstract In this paper we describe the design and implementation of a distributed operating system for sensor net- works. The goal of our system is to extend total system lifetime through power-aware adaptation for sensor networking applications. Our system achieves this goal by providing a single system image of a unified Java virtual machine to applications over an ad hoc collection of heterogeneous sensors. It automatically and transparently partitions applications into components and dynamically finds a placement of these components on nodes within the sensor network to reduce energy consumption and increase system longevity. This paper describes the design and implementation of our system and examines the question of where and when to mi- grate components in a sensor network. We evaluate two practical, power-aware, general-purpose algorithms for object placement, as well as an adaptive scheme for deciding the time granularity of object migration. We demonstrate that our algorithms can increase sensor network longevity by a factor of four to five by effec- tively distributing energy consumption and avoiding hotspots. 1. Introduction able to components at each node, in particular the available power and bandwidth may change over Sensor networks simultaneously promise a radi- time and necessitate the relocation of application cally new class of applications and pose signifi- components. Further, event sources that are being cant challenges for application development.
    [Show full text]
  • Load Balancing Experiments in Openmosix”, Inter- National Conference on Computers and Their Appli- Cations , Seattle WA, March 2006 0 0 5 4 9
    Machine Learning Approach to Tuning Distributed Operating System Load Balancing Algorithms Dr. J. Michael Meehan Alan Ritter Computer Science Department Computer Science Department Western Washington University Western Washington University Bellingham, Washington, 98225 USA Bellingham, Washington, 98225 USA [email protected] [email protected] Abstract 2.6 Linux kernel openMOSIX patch was in the process of being developed. This work concerns the use of machine learning techniques (genetic algorithms) to optimize load 2.1 Load Balancing in openMosix balancing policies in the openMosix distributed The standard load balancing policy for openMosix operating system. Parameters/alternative algorithms uses a probabilistic, decentralized approach to in the openMosix kernel were dynamically disseminate load balancing information to other altered/selected based on the results of a genetic nodes in the cluster. [4] This allows load information algorithm fitness function. In this fashion optimal to be distributed efficiently while providing excellent parameter settings and algorithms choices were scalability. The major components of the openMosix sought for the loading scenarios used as the test load balancing scheme are the information cases. dissemination and migration kernel daemons. The information dissemination daemon runs on each 1 Introduction node and is responsible for sending and receiving The objective of this work is to discover ways to load messages to/from other nodes. The migration improve the openMosix load balancing algorithm. daemon receives migration requests from other The approach we have taken is to create entries in nodes and, if willing, carries out the migration of the the /proc files which can be used to set parameter process. Thus, the system is sender initiated to values and select alternative algorithms used in the offload excess load.
    [Show full text]
  • HPC with Openmosix
    HPC with openMosix Ninan Sajeeth Philip St. Thomas College Kozhencheri IMSc -Jan 2005 [email protected] Acknowledgements ● This document uses slides and image clippings available on the web and in books on HPC. Credit is due to their original designers! IMSc -Jan 2005 [email protected] Overview ● Mosix to openMosix ● Why openMosix? ● Design Concepts ● Advantages ● Limitations IMSc -Jan 2005 [email protected] The Scenario ● We have MPI and it's pretty cool, then why we need another solution? ● Well, MPI is a specification for cluster communication and is not a solution. ● Two types of bottlenecks exists in HPC - hardware and software (OS) level. IMSc -Jan 2005 [email protected] Hardware limitations for HPC IMSc -Jan 2005 [email protected] The Scenario ● We are approaching the speed and size limits of the electronics ● Major share of possible optimization remains with software part - OS level IMSc -Jan 2005 [email protected] Hardware limitations for HPC IMSc -Jan 2005 [email protected] How Clusters Work? Conventional supe rcomputers achieve their speed using extremely optimized hardware that operates at very high speed. Then, how do the clusters out-perform them? Simple, they cheat. While the supercomputer is optimized in hardware, the cluster is so in software. The cluster breaks down a problem in a special way so that it can distribute all the little pieces to its constituents. That way the overall problem gets solved very efficiently. - A Brief Introduction To Commodity Clustering Ryan Kaulakis IMSc -Jan 2005 [email protected] What is MOSIX? ● MOSIX is a software solution to minimise OS level bottlenecks - originally designed to improve performance of MPI and PVM on cluster systems http://www.mosix.org Not open source Free for personal and academic use IMSc -Jan 2005 [email protected] MOSIX More Technically speaking: ● MOSIX is a Single System Image (SSI) cluster that allows Automated Load Balancing across nodes through preemptive process migrations.
    [Show full text]
  • Introduction Course Outline Why Does This Fail? Lectures Tutorials
    Course Outline • Prerequisites – COMP2011 Data Organisation • Stacks, queues, hash tables, lists, trees, heaps,…. Introduction – COMP2021 Digital Systems Structure • Assembly programming • Mapping of high-level procedural language to assembly COMP3231/9201/3891/9283 language – or the postgraduate equivalent (Extended) Operating Systems – You are expected to be competent Kevin Elphinstone programmers!!!! • We will be using the C programming language – The dominant language for OS implementation. – Need to understand pointers, pointer arithmetic, explicit 2 memory allocation. Why does this fail? Lectures • Common for all courses (3231/3891/9201/9283) void func(int *x, int *y) • Wednesday, 2-4pm { • Thursday, 5-6pm *x = 1; *y = 2; – All lectures are here (EE LG03) – The lecture notes will be available on the course web site } • Available prior to lectures, when possible. void main() – The lecture notes and textbook are NOT a substitute for { attending lectures. int *a, *b; func(a,b); } 3 4 Tutorials Assignments • Assignments form a substantial component of • Start in week 2 your assessment. • A tutorial participation mark will • They are challenging!!!! – Because operating systems are challenging contribute to your final assessment. • We will be using OS/161, – Participation means participation, NOT – an educational operating system attendance. – developed by the Systems Group At Harvard – Comp9201/3891/9283 students excluded – It contains roughly 20,000 lines of code and comments • You will only get participation marks in your enrolled tutorial. 5 6 1 Assignments Assignments • Assignments are in pairs • Don’t under estimate the time needed to do the – Info on how to pair up available soon assignments. • We usually offer advanced versions of the – ProfQuotes: [About the midterm] "We can't keep you working assignments on it all night, it's not OS.“ Ragde, CS341 – Available bonus marks are small compared to amount of • If you start a couple days before they are due, you will be late.
    [Show full text]
  • The Utility Coprocessor: Massively Parallel Computation from the Coffee Shop
    The Utility Coprocessor: Massively Parallel Computation from the Coffee Shop John R. Douceur, Jeremy Elson, Jon Howell, and Jacob R. Lorch Microsoft Research Abstract slow jobs that take several minutes without UCop be- UCop, the “utility coprocessor,” is middleware that come interactive (15–20 seconds) with it. Thanks to makes it cheap and easy to achieve dramatic speedups the recent emergence of utility-computing services like of parallelizable, CPU-bound desktop applications using Amazon EC2 [8] and FlexiScale [45], which rent com- utility computing clusters in the cloud. To make UCop puters by the hour on a moment’s notice, anyone with a performant, we introduced techniques to overcome the credit card and $10 can use UCop to speed up his own low available bandwidth and high latency typical of the parallel applications. networks that separate users’ desktops from a utility One way to describe UCop is that it effectively con- computing service. To make UCop economical and easy verts application software into a scalable cloud service to use, we devised a scheme that hides the heterogene- targeted at exactly one user. This goal entails five re- ity of client configurations, allowing a single cluster to quirements. Configuration transparency means the ser- serve virtually everyone: in our Linux-based prototype, vice matches the user’s application, library, and con- the only requirement is that users and the cluster are us- figuration state. Non-invasive installation means UCop ing the same major kernel version. works with a user’s existing file system and application This paper presents the design, implementation, and configuration.
    [Show full text]
  • Legoos: a Disseminated, Distributed OS for Hardware Resource
    LegoOS: A Disseminated, Distributed OS for Hardware Resource Disaggregation Yizhou Shan, Yutong Huang, Yilun Chen, and Yiying Zhang, Purdue University https://www.usenix.org/conference/osdi18/presentation/shan This paper is included in the Proceedings of the 13th USENIX Symposium on Operating Systems Design and Implementation (OSDI ’18). October 8–10, 2018 • Carlsbad, CA, USA ISBN 978-1-939133-08-3 Open access to the Proceedings of the 13th USENIX Symposium on Operating Systems Design and Implementation is sponsored by USENIX. LegoOS: A Disseminated, Distributed OS for Hardware Resource Disaggregation Yizhou Shan, Yutong Huang, Yilun Chen, Yiying Zhang Purdue University Abstract that can fit into monolithic servers and deploying them in datacenters is a painful and cost-ineffective process that The monolithic server model where a server is the unit often limits the speed of new hardware adoption. of deployment, operation, and failure is meeting its lim- We believe that datacenters should break mono- its in the face of several recent hardware and application lithic servers and organize hardware devices like CPU, trends. To improve resource utilization, elasticity, het- DRAM, and disks as independent, failure-isolated, erogeneity, and failure handling in datacenters, we be- network-attached components, each having its own con- lieve that datacenters should break monolithic servers troller to manage its hardware. This hardware re- into disaggregated, network-attached hardware compo- source disaggregation architecture is enabled by recent nents. Despite the promising benefits of hardware re- advances in network technologies [24, 42, 52, 66, 81, 88] source disaggregation, no existing OSes or software sys- and the trend towards increasing processing power in tems can properly manage it.
    [Show full text]
  • Architectural Review of Load Balancing Single System Image
    Journal of Computer Science 4 (9): 752-761, 2008 ISSN 1549-3636 © 2008 Science Publications Architectural Review of Load Balancing Single System Image Bestoun S. Ahmed, Khairulmizam Samsudin and Abdul Rahman Ramli Department of Computer and Communication Systems Engineering, University Putra Malaysia, 43400 Serdang, Selangor, Malaysia Abstract: Problem statement: With the growing popularity of clustering application combined with apparent usability, the single system image is in the limelight and actively studied as an alternative solution for computational intensive applications as well as the platform for next evolutionary grid computing era. Approach: Existing researches in this field concentrated on various features of Single System Images like file system and memory management. However, an important design consideration for this environment is load allocation and balancing that is usually handled by an automatic process migration daemon. Literature shows that the design concepts and factors that affect the load balancing feature in an SSI system are not clear. Result: This study will review some of the most popular architecture and algorithms used in load balancing single system image. Various implementations from the past to present will be presented while focusing on the factors that affect the performance of such system. Conclusion: The study showed that although there are some successful open source systems, the wide range of implemented systems investigated that research activity should concentrate on the systems that have already been proposed and proved effectiveness to achieve a high quality load balancing system. Key words: Single system image, NOWs (network of workstations), load balancing algorithm, distributed systems, openMosix, MOSIX INTRODUCTION resources transparently irrespective of where they are available[1].The load balancing single system image Cluster of computers has become an efficient clusters dominate research work in this environment.
    [Show full text]