DOCSLIB.ORG

Apache/Mesos

Total Page：16

File Type：pdf, Size：1020Kb

Download full-text PDF Read full-text
Apache/Mesos Re:platforming// the/Datacenter// with/Apache/Mesos/ Christos(Kozyrakis( Why$your$ASF$project$should$run$on$Mesos$ O(10K)/commodity/servers/ HighAspeed/networking/ Distributed/storage/(HDD,/Flash)/ x10/MWatt/ x100/M$/ ( developers$ ops$ automation( automation( performance( efficiency( ( ( ① Datacenter/past/ Static/Partitioning/ Static/Partitioning/ Hadoop/ Cassandra/ Rails/ Jenkins/ memcached/ Static/Partitioning/ Hadoop/ Cassandra/ Rails/ Jenkins/ memcached/ Static/Partitioning/ Hadoop/ Cassandra/ Rails/ Jenkins/ memcached/ Static/Partitioning/ Hadoop/ Cassandra/ Rails/ Jenkins/ memcached/ Static/Partitioning/ developers$ ops$ " automation( " automation( ! performance( " efficiency( ( ( ② Datacenter/present/ Apache/Mesos/ The(datacenter(OS(kernel( Aggregates(all(resources(into(a(single(shared(pool( Dynamically(allocates(resources(to(distributed(apps( Container(management(at(scale((cgroups,(docker,(…)( Mesos/Architecture/ Frameworks/ Marathon( Jenkins( … Allocator/ Allocator/ Allocator/ Masters/ AuthN/ AuthZ/ AuthN/ AuthZ/ AuthN/ AuthZ/ Slave/ Slave/ Executor/ Executor/ … Executor/ Executor/ Servers/ Task/ Task/ Task/ Task/ Scales(to(10s(of(thousands(of(servers( Mesos/Fault/Tolerance/ Frameworks/ Marathon( Jenkins( … Allocator/ Allocator/ Allocator/ Masters/ AuthN/ AuthZ/ AuthN/ AuthZ/ AuthN/ AuthZ/ Slave/ Slave/ Executor/ Executor/ … Executor/ Executor/ Servers/ Task/ Task/ Task/ Task/ Tasks(survive(failures(of(the(master((( Mesos/Fault/Tolerance/ Frameworks/ Marathon( Jenkins( … Allocator/ Allocator/ Allocator/ Masters/ AuthN/ AuthZ/ AuthN/ AuthZ/ AuthN/ AuthZ/ Slave/ Slave/ Executor/ Executor/ … Executor/ Executor/ Servers/ Task/ Task/ Task/ Task/ Tasks(survive(failures(of(the(framework( Mesos/Fault/Tolerance/ Frameworks/ Marathon( Jenkins( … Allocator/ Allocator/ Allocator/ Masters/ AuthN/ AuthZ/ AuthN/ AuthZ/ AuthN/ AuthZ/ Slave/ Slave/ Executor/ Executor/ … Executor/ Executor/ Servers/ Task/ Task/ Task/ Task/ Tasks(survive(failures(of(the(slave(process( Scheduling/in/Mesos/ No(single(scheduler(fits(all(needs( LongLrunning(services(need( (scale(up/down,(fault(tolerance( ( Analytics(services(need( (fast(task(launching( 2Alevel/Scheduling/ Mesos(master((single(API)( Resource(allocation((offers)(( Task(health(checks( Task(isolation( Mesos(frameworks((domainLspecific(APIs)( Scale(up/down( Fault(tolerance( Task(grouping( Task(dependencies( Queuing(&(priorities(( 2Alevel/Scheduling/Benefits/ Multiple(APIs(to(Mesos(through(frameworks( Marathon,(Aurora,(Singularity( Storm,(Spark,(Hadoop,(Chronos( Cassandra,(Elasticsearch( Multiple(task(scheduling(approaches( Spark(fineLgrain(Vs(Spark(coarseLgrain( ( Simple,(stable,(and(scalable(Mesos(master( Dynamic/Resource/Allocation// Resource(allocation(based(on(framework(roles( ( Dominant(resource(fairness((DRF)( Weighted(fair(share(calculated(based(on(dominant(resource( Frameworks(do(no(worse(than(having(a(weightLsized(cluster( ( Resource(reservations( Resources(can(be(allocated(to(specific(frameworks(if(needed( ( Dynamic/Resource/Allocation// Rails/ Hadoop/ buy$less$machines$ memcached/ or$ run$more$applications!$ Service/Discovery/ ① Watch(ZK(for( ((master(changes( Mesos( Mesos( Master( ② Pull(task(state( DNS( (Generate(DNS(records( ③ DNS(&(HTTP( (based(discovery( Slave( Slave( Slave( Slave( … Slave( (nginx.marathon.mesos(#(10.13.17.95( _nginx._tcp.marathon.mesos(#10.13.17.95:8181( ( ③ Datacenter/future/ Utilization/Reality/ Twitter (Mesos) Google (Borg) [Delimitrou’14] [Barroso’09] The/Curse/of/Overprovisioning/ [Delimitrou’14] Bloated(reservations(to(deal(with( diurnal(load(patterns,(load(spikes,(software(&(platform(changes( Oversubscription/ Frameworks/ Marathon( Jenkins( … ② offer<s1,(4cores,(…>( Allocator/ Allocator/ Allocator/ Masters/ AuthN/ AuthZ/ AuthN/ AuthZ/ AuthN/ AuthZ/ ① offer<s1,(4cores,(…>( Slave/ Slave/ Executor/ Executor/ . Executor/ Executor/ Servers/ Task/ Task/ Task/ Task/ Oversubscription/ Frameworks/ Marathon( Jenkins( … ③ launch<tasks,(s1,(4cores,(…>( Allocator/ Allocator/ Allocator/ Masters/ AuthN/ AuthZ/ AuthN/ AuthZ/ AuthN/ AuthZ/ ④ launch<tasks,(4cores,(…>( Slave/ Slave/ Executor/ Executor/ . Executor/ Executor/ Servers/ Task/ Task/ Task/ Task/ Oversubscription/ Frameworks/ Marathon( Jenkins( … ② offer<s1,(BE,(2cores,(…>( Allocator/ Allocator/ Allocator/ Masters/ AuthN/ AuthZ/ AuthN/ AuthZ/ AuthN/ AuthZ/ ① offer<s1,(BE,2cores,(…>( Slave/ Slave/ Executor/ Executor/ . Executor/ Executor/ Servers/ Task/ Task/ Task/ Task/ Oversubscription/ Frameworks/ Marathon( Jenkins( … ③ launch<tasks,(BE,(s1,(2cores,(…>( Allocator/ Allocator/ Allocator/ Masters/ AuthN/ AuthZ/ AuthN/ AuthZ/ AuthN/ AuthZ/ ④ launch<tasks,(BE,(2cores,(…>( Slave/ Slave/ Executor/ Executor/ . Executor/ Executor/ Servers/ Task/ Task/ Task/ Task/ Oversubscription/ Frameworks/ Marathon( Jenkins( … ② Status<task,(killed,(…>( Allocator/ Allocator/ Allocator/ Masters/ AuthN/ AuthZ/ AuthN/ AuthZ/ AuthN/ AuthZ/ ① Status<task,(killed,(…>( Slave/ Slave/ Executor/ Executor/ . Executor/ Executor/ Servers/ Task/ Task/ Task/ Task/ Interference/#/Performance/Loss/ Impact of interference on websearch’s latency 300% L3 Cache >300%( >300%( >300%( >300%( >300%( >300%( >300%( 264%( 123%( DRAM >300%( >300%( >300%( >300%( >300%( >300%( >300%( 270%( 122%( B A D HyperThread 110%( 107%( 114%( 115%( 105%( 117%( 120%( 136%( >300%( CPU power 124%( 107%( 116%( 109%( 115%( 105%( 101%( 100%( 100%( 100% O K Network 36%( 36%( 37%( 37%( 39%( 42%( 48%( 55%( 64%( 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% [Lo’15] Load Isolators/ Mesos(slave(invokes(isolators( Modules(that(monitor(&(isolate(resources(for(executors( Isolator(modules( CPU((cgroups(cpushares,(cpusets)( Memory((cgroups)( Disk( Network( Cache( Power( …(( Isolators/#/Performance/QoS/ [Lo et al’15] +/bestAeffort/task/ >90%/HW/utilization/ No/latency/SLO/problems/ Hybrid/Datacenters/ Marathon( Mesos/Master/ Mesos/Slaves/ Hybrid/Datacenters/ Marathon( ?/ Mesos/Master/ instance(++( Mesos/Slaves/ Scheduling(based(on(workload(type,(data(locality,(pricing,…(( Future/Directions/ Oversubscription( ((( Container(&(application(right(sizing( ((( Hybrid(datacenters( ((( Power(aware(scheduling( ((( Locality(aware(scheduling( Mesos/Datacenter/ developers$ ops$ ! automation( ! automation( ! performance( ! efficiency( ( ( Want/to/Learn/More?/ ( Mo(3pm(–(Cracking(the(Container(Scale(Problem(with(Apache(Mesos( ( Tue(4.20pm(–(The(Emergence(of(the(Datacenter(Developer( (( Wed(4.15pm(–(Mesos(+(Yarn(=(Myriad( Questions?/ ( ( (( References/ • http://mesos.apache.org/( • http://www.mesosphere.com/( • https://github.com/mesosphere/mesosLdns( • https://www.cs.berkeley.edu/~alig/papers/mesos.pdf( • https://www.cs.berkeley.edu/~alig/papers/drf.pdf( • http://www.morganclaypool.com/doi/abs/10.2200/ S00516ED2V01Y201306CAC024( • http://web.stanford.edu/~cdel/2014.asplos.quasar.pdf( • http://web.stanford.edu/~davidlo/resources/2014.heracles.isca.pdf(.
Load more

Recommended publications
  • Myriad: Resource Sharing Beyond Boundaries
    Resource Sharing Beyond Boundaries Mohit Soni Santosh Marella Adam Bordelon Anoop Dawar Ben Hindman Brandon Gulla Danese Cooper Darin Johnson Jim Klucar Kannan Rajah Ken Sipe Luciano Resende Meghdoot Bhattacharya Paul Reed Renan DelValle Ruth Harris Shingo Omura Swapnil Daingade Ted Dunning Will Ochandarena Yuliya Feldman Zhongyue Luo Agenda What's up with Datacenters these days? Apache Mesos vs. Apache Hadoop/YARN? Why would you want/need both? Resource Sharing with Apache Myriad What's running on your datacenter? Tier 1 services Tier 2 services High Priority Batch Best Effort, backfill Requirements Programming models based on resources, not machines Custom resource types Custom scheduling algorithms: Fast vs. careful/slow Lightweight executors, fast task launch time Multi-tenancy, utilization, strong isolation Hadoop and More Support Hadoop/BigData ecosystem Support arbitrary (legacy) processes/containers Connect Big Data to non-Hadoop apps, share data, resources Mesos from 10,000 feet Open Source Apache project Cluster Resource Manager Scalable to 10,000s of nodes Fault-tolerant, no SPOF Multi-tenancy, Resource Isolation Improved resource utilization Mesos is more than Yet Another Resource Negotiator Long-running services; real-time jobs Native Docker; cgroups for years; Isolate cpu/mem/disk/net/other Distributed systems SDK; ~200 loc for a new app Core written in C++ for performance, Apps in any language Why two resource managers? Static Partitioning sucks Hadoop teams fine with isolated clusters, but Ops team unhappy; slow
    [Show full text]
  • Is 'Distributed' Worth It? Benchmarking Apache Spark with Mesos
    Is `Distributed' worth it? Benchmarking Apache Spark with Mesos N. Satra (ns532) January 13, 2015 Abstract A lot of research focus lately has been on building bigger dis- tributed systems to handle `Big Data' problems. This paper exam- ines whether typical problems for web-scale companies really benefits from the parallelism offered by these systems, or can be handled by a single machine without synchronisation overheads. Repeated runs of a movie review sentiment analysis task in Apache Spark were carried out using Apache Mesos and Mesosphere, on Google Compute Engine clusters of various sizes. Only a marginal improvement in run-time was observed on a distributed system as opposed to a single node. 1 Introduction Research in high performance computing and `big data' has recently focussed on distributed computing. The reason is three-pronged: • The rapidly decreasing cost of commodity machines, compared to the slower decline in prices of high performance or supercomputers. • The availability of cloud environments like Amazon EC2 or Google Compute Engine that let users access large numbers of computers on- demand, without upfront investment. • The development of frameworks and tools that provide easy-to-use id- ioms for distributed computing and managing such large clusters of machines. Large corporations like Google or Yahoo are working on petabyte-scale problems which simply can't be handled by single computers [9]. However, 1 smaller companies and research teams with much more manageable sizes of data have jumped on the bandwagon, using the tools built by the larger com- panies, without always analysing the performance tradeoffs. It has reached the stage where researchers are suggesting using the same MapReduce idiom hammer on all problems, whether they are nails or not [7].
    [Show full text]
  • The Dzone Guide to Volume Ii
    THE D ZONE GUIDE TO MODERN JAVA VOLUME II BROUGHT TO YOU IN PARTNERSHIP WITH DZONE.COM/GUIDES DZONE’S 2016 GUIDE TO MODERN JAVA Dear Reader, TABLE OF CONTENTS 3 EXECUTIVE SUMMARY Why isn’t Java dead after more than two decades? A few guesses: Java is (still) uniquely portable, readable to 4 KEY RESEARCH FINDINGS fresh eyes, constantly improving its automatic memory management, provides good full-stack support for high- 10 THE JAVA 8 API DESIGN PRINCIPLES load web services, and enjoys a diverse and enthusiastic BY PER MINBORG community, mature toolchain, and vigorous dependency 13 PROJECT JIGSAW IS COMING ecosystem. BY NICOLAI PARLOG Java is growing with us, and we’re growing with Java. Java 18 REACTIVE MICROSERVICES: DRIVING APPLICATION 8 just expanded our programming paradigm horizons (add MODERNIZATION EFFORTS Church and Curry to Kay and Gosling) and we’re still learning BY MARKUS EISELE how to mix functional and object-oriented code. Early next 21 CHECKLIST: 7 HABITS OF SUPER PRODUCTIVE JAVA DEVELOPERS year Java 9 will add a wealth of bigger-picture upgrades. 22 THE ELEMENTS OF MODERN JAVA STYLE But Java remains vibrant for many more reasons than the BY MICHAEL TOFINETTI robustness of the language and the comprehensiveness of the platform. JVM languages keep multiplying (Kotlin went 28 12 FACTORS AND BEYOND IN JAVA GA this year!), Android keeps increasing market share, and BY PIETER HUMPHREY AND MARK HECKLER demand for Java developers (measuring by both new job 31 DIVING DEEPER INTO JAVA DEVELOPMENT posting frequency and average salary) remains high. The key to the modernization of Java is not a laundry-list of JSRs, but 34 INFOGRAPHIC: JAVA'S IMPACT ON THE MODERN WORLD rather the energy of the Java developer community at large.
    [Show full text]
  • Kubernetes As an Availability Manager for Microservice Based Applications Leila Abdollahi Vayghan
    Kubernetes as an Availability Manager for Microservice Based Applications Leila Abdollahi Vayghan A Thesis in the Department of Computer Science and Software Engineering Presented in Partial Fulfillment of the Requirements for the Degree of Master of Computer Science at Concordia University Montreal, Quebec, Canada August 2019 © Leila Abdollahi Vayghan, 2019 CONCORDIA UNIVERSITY SCHOOL OF GRADUATE STUDIES This is to certify that the thesis prepared By: Leila Abdollahi Vayghan Entitled: Kubernetes as an Availability Manager for Microservice Based Applications and submitted in partial fulfillment of the requirements for the degree of Master in Computer Science complies with the regulations of the University and meets the accepted standards with respect to originality and quality. Signed by the final examining committee: ________________________________________________ Chair Dr. P. Rigby ________________________________________________ Internal Examiner Dr. D. Goswami ________________________________________________ Internal Examiner Dr. J. Rilling ________________________________________________ Co-Supervisor Dr. F. Khendek ________________________________________________ Co-Supervisor Dr. M. Toeroe Approved by: ___________________________________ Dr. L. Narayanan, Chair Department of Computer Science and Software Engineering _______________ 2019___ __________________________________ Dr. Amir Asif, Dean, Faculty of Engineering and Computer Science ii ABSTRACT Kubernetes as an Availability Manager for Microservice Based Applications Leila
    [Show full text]
  • A Single Platform for Container Orchestration and Data Services
    A SINGLE PLATFORM FOR CONTAINER ORCHESTRATION AND DATA SERVICES MESOSPHERE DC/OS WITH KUBERNETES EASES ENTERPRISE ADOPTION OF NEW TECHNOLOGIES FOR DIGITAL TRANSFORMATION EXECUTIVE SUMMARY Digital disruption is occurring across many industries as technology improvements, ranging from mobile and social to Internet of Things (IoT), are shifting customer behavior and market context. Enterprises mastering digital business technologies are positioned to take advantage of this trend, while poorly equipped incumbents are left behind. This is so widely recognized that achieving a faster pace of innovation is commonly considered a mission-critical investment. Achieving the goals of the investment depends on an organization’s ability to rapidly: ● Iterate in application development – continuously delivering improvements, ● Extract value from data – focused on immediate decision making and response with “fast data” as well as in-depth analytical investigation with “big data”. Best practice rapid application development requires the use of tools like Kubernetes, which serves as a container orchestrator supporting cloud-native methods of continuous deployment, DevOps, and microservices architecture for running stateless applications such as web and application servers. To maximize value from an organization’s data in determining insights and actions requires another set of tools from a diverse, evolving set of stateful (i.e. data store) applications like Spark, Cassandra, and Kafka. Open source tools and cloud computing are a great start, but their rate of change and complexity to operate (i.e. upgrade, scale, etc.) are challenging for most organizations to embrace. Organizations commonly seek a platform integrating a set of tools into a solution. The struggle has been to find a platform that addresses ease of running both cloud-native, stateless applications and data-intensive, stateful applications.
    [Show full text]
  • Delivering Business Value with Apache Mesos
    White paper: Delivering Business Value with Apache Mesos Executive Summary Case Studies In today’s business environment, time to market is critical as we eBay increased developer productivity and reduced operational are more reliant on technology to meet customer needs. costs associated with virtualisation machine sprawl at eBay Traditional approaches to solving technology problems are failing as we develop and run applications at massive scale. eBay Inc - The worlds leading online auction site Apache Mesos is an open source project that addresses the eBay have heavily invested in continuous integration as a problems of efciently managing data center infrastructure at development best practice. At eBay, each developer gets a scale. Apache Mesos also provides a simplifed abstraction layer dedicated virtual machine running Jenkins, a widely used for your development resources to use, allowing them to focus on continuous integration server. This resulted in 1000’s of VMs delivering customer value. (Virtual Machines), which used physical resources and required operational support. In this white paper, we look at some of the core business drivers in the digital age, the problems of running and developing technology at scale and examples from disruptive companies like Continuous Integration is a development Ebay, Twitter and Airb&B who have Mesos in production. practice that requires developers to The Problems of Technology at Scale integrate code into a shared repository several times a day. Each check-in is then Development teams are spending more time dealing with increasingly complex infrastructures, rather than delivering on verified by an automated build, allowing new product features. Computing resources in our data centers teams to detect problems early.
    [Show full text]
  • Guide to the Open Cloud Open Cloud Projects Profiled
    Guide to the Open Cloud Open cloud projects profiled A Linux Foundation publication January 2015 www.linuxfoundation.org Introduction The open source cloud computing landscape has changed significantly since we published our first cloud guide in October 2013. This revised version adds new projects See also the rise of Linux container and technology categories that have since technology with the advent of Docker gained importance, and in some cases and its emerging ecosystem. You will be radically change how companies approach hard pressed to find an enterprise Linux building and deploying an open source distribution that isn’t yet working on Docker cloud architecture. integration and touting its new container strategy. Even VMware vSphere, Google In 2013, many cloud projects were still Cloud Platform, and Microsoft Azure are working out their core enterprise features rushing to adapt their cloud platforms to the and furiously building in functionality. And open source Docker project. enterprises were still very much in the early stages of planning and testing their public, This rapid pace of innovation and resulting private or hybrid clouds–and largely at the disruption of existing platforms and vendors orchestration layer. can now serve as a solid case study for the role of open source software and Now, not only have cloud projects collaboration in advancing the cloud. consistently (and sometimes dramatically) grown their user and developer Other components of the cloud infrastructure communities, lines of code and commits have also followed suit, hoping to harness over the past year, their software is the power of collaboration. The Linux increasingly enterprise-ready.
    [Show full text]
  • Two Stage Cluster for Resource Optimization with Apache Mesos
    Two stage cluster for resource optimization with Apache Mesos Gourav Rattihalli1, Pankaj Saha1, Madhusudhan Govindaraju1, and Devesh Tiwari2 1Cloud and Big Data Lab, State University of New York (SUNY) at Binghamton 2Northeastern University fgrattih1, psaha4, [email protected] and [email protected] Abstract—As resource estimation for jobs is difficult, users At Twitter, their large-scale cluster uses Mesos for resource often overestimate their requirements. Both commercial clouds management. This cluster with thousands of server class nodes and academic campus clusters suffer from low resource utiliza- has reservations reaching 80% but the utilization has been tion and long wait times as the resource estimates for jobs, provided by users, is inaccurate. We present an approach to found be to consistently below 20% [3]. statistically estimate the actual resource requirement of a job At SUNY Binghamton’s Spiedie campus cluster, the snap- in a Little cluster before the run in a Big cluster. The initial shot of a single day’s data shows that overall users requested estimation on the little cluster gives us a view of how much significantly more CPU resources than required - the number actual resources a job requires. This initial estimate allows us to of cores requested was 7975 but the actual usage was 4415 accurately allocate resources for the pending jobs in the queue and thereby improve throughput and resource utilization. In our cores. experiments, we determined resource utilization estimates with When commercial cloud facilities such as Amazon EC2 an average accuracy of 90% for memory and 94% for CPU, and Microsoft Azure operate at very low resource utilization, while we make better utilization of memory by an average of it increases the cost to operate the facilities, and thereby 22% and CPU by 53%, compared to the default job submission the cost for end users.
    [Show full text]
  • 2017 Kevin Klues [email protected]
    NVIDIA GPU Support for Apache Mesos and DC/OS GPU Technology Conference - 2017 Kevin Klues [email protected] © 2017 Mesosphere, Inc. All Rights Reserved. 1 Kevin Klues is a Tech Lead Manager at Mesosphere working with both the Mesos core team as well as the DC/OS Provisioning and Management team. Since joining Mesosphere, Kevin has been involved in the design and implementation of a number of Mesos’s core subsystems, including GPU isolation, Pods, and Attach/Exec support. Prior to joining Mesosphere, Kevin worked at Google on an experimental operating system for data centers called Akaros. He and a few others founded the Akaros project while working on their Ph.Ds at UC Berkeley. In a past life Kevin was a lead developer of the TinyOS project, working at Stanford, the Technical University of Berlin, and the CSIRO in Australia. When not working, you can usually find Kevin on a snowboard or up in the mountains in some capacity or another. © 2017 Mesosphere, Inc. All Rights Reserved. 2 What is Apache Mesos? ● An open-source, distributed systems kernel (a.k.a cluster manager) for fine-grained management of cluster resources and tasks © 2017 Mesosphere, Inc. All Rights Reserved. 3 What is Apache Mesos? © 2017 Mesosphere, Inc. All Rights Reserved. 4 What is Apache Mesos? ● Mesos provides its own containerization technology (called the Mesos containerizer) ● It supports the standard docker image format, but relies on its own internal implementation for building containers ● A separate docker containerizer is also available, but not relevant to this presentation © 2017 Mesosphere, Inc. All Rights Reserved.
    [Show full text]
  • Deploying Apache Flink at Scale
    FlinkForward 2017 - San Francisco Flink meet DC/OS Deploying Apache Flink at Scale Elizabeth K. Joseph, @pleia2 Ravi Yadav, @RaaveYadav © 2017 Mesosphere, Inc. All Rights Reserved. 1 Talk Outline Part 1 Part 2 Part 3 Introduction to Apache Demonstration of demo DC/OS 1.9 key features for Mesos, Marathon, and data pipeline + Installing data services and beyond DC/OS Flink on DC/OS © 2017 Mesosphere, Inc. All Rights Reserved. 2 Apache Mesos: The datacenter kernel http://mesos.apache.org/ © 2017 Mesosphere, Inc. All Rights Reserved. 3 Marathon ● Mesos can’t run applications on its own. ● A Mesos framework is a distributed system that has a scheduler. ● Schedulers like Marathon start and keep your applications running. A bit like a distributed init system. ● Mesos mechanics are fair and HA ● Learn more at https://mesosphere.github.io/marat hon/ © 2017 Mesosphere, Inc. All Rights Reserved. 4 Introducing DC/OS Solves common problems ● Resource management ● Task scheduling ● Container orchestration ● Self-healing infrastructure ● Logging and metrics ● Network management ● “Universe” of pre-configured apps (including Flink, Kafka…) ● Learn more and contribute at https://dcos.io/ © 2017 Mesosphere, Inc. All Rights Reserved. 5 Services & Containers DC/OS HDFS Jenkins Marathon Cassandra Flink Architecture Overview Spark Docker Kafka MongoDB +30 more... DC/OS Security & Container Orchestration Monitoring & Operations User Interface & Command Line Governance ANY INFRASTRUCTURE © 2017 Mesosphere, Inc. All Rights Reserved. 6 Interact with DC/OS (1/2) Web-based GUI https://dcos.io/docs/lates t/usage/webinterface/ © 2017 Mesosphere, Inc. All Rights Reserved. 7 Universe © 2017 Mesosphere, Inc. All Rights Reserved. 8 Interact with DC/OS (2/2) CLI tool API https://dcos.io/docs/latest/usage/cli/ https://dcos.io/docs/latest/api/ © 2017 Mesosphere, Inc.
    [Show full text]
  • Why Kubernetes Matters
    WHITE PAPER Why Kubernetes Matters We are on the cusp of a new era of software: instead of bolting on operations as an afterthought to the software development process, Kubernetes promises to bring development and operations together by design. With declarative, infrastructure- agnostic constructs to describe how applications are composed, interact and managed, Kubernetes enables an order of magnitude increase in operability of modern software systems. This document describes why Kubernetes matters, and why it represents a significant step forward for DevOps organizations. WHITE PAPER: Why Kubernetes Introduction Matters Over the last 2-3 years, Docker containers have made it very easy to run cloud-native applications on physical or virtual infrastructure. They are lighter weight compared to VMs and make more efficient use of the underlying infrastruc- ture. Containers are meant to make it easy to turn apps on and off to meet fluctuating demand and move applications seamlessly between different environments e.g. Dev/Test/Production, or even across clouds. While the container runtime APIs meet the needs of managing an individual container on one host, they are not suited to managing applications running in multiple containers deployed across multiple hosts. This is where container orchestration tools, such as Kubernetes, are required. According to the Kubernetes website – “Kubernetes is an open-source system for automating deployment, scaling, and management of containerized applications.” Kubernetes was built by Google based on their experience running containers in production over the last decade. Kubernetes Benefits Write once, run anywhere (any cloud) DevOps teams today must work across a wide range of infrastructure choices: on-premises dedicated servers, on-premises private cloud (IaaS), public clouds such as AWS, GCP and Azure, or a hybrid cloud across these.
    [Show full text]
  • Fall 2020 Vol
    ;login FALL 2020 VOL. 45, NO. 3 : & Guide to the Performance of Optane DRAM Jian Yang, Juno Kim, Morteza Hoseinzadeh, Joseph Izraelevitz, and Steven Swanson & Understanding Linux Superpages Weixi Zhu, Alan L. Cox, and Scott Rixner & Interview with Ion Stoica about Serverless Computing Rik Farrow & Measuring the Health of Open Source Projects Georg J.P. Link & Using OpenTrace to Track DB Performance Problems Anatoly Mikhaylov Columns SRE and Justice Laura Nolan Socially Distant Projects Cory Lueninghoener Communicating with BPF Dave Josephsen Hand-Over-Hand Locking for Highly Concurrent Collections Terence Kelly Everything Is a Punch Card Simson L. Garfinkel Thanks to our USENIX Supporters! USENIX appreciates the financial assistance our Supporters provide to subsidize our day-to-day operations and to continue our non-profit mission. Our supporters help ensure: • Free and open access to technical information • Student Grants and Diversity Grants to participate in USENIX conferences • The nexus between academic research and industry practice • Diversity and representation in the technical workplace We need you now more than ever! Contact us at [email protected]. USENIX PATRONS USENIX BENEFACTORS We offer our heartfelt appreciation to the following sponsors and champions of conference diversity, open access, and our SREcon communities via their sponsorship of multiple conferences: Ethyca Dropbox Microsoft Azure Packet Datadog Goldman Sachs LinkedIn Salesforce More information at www.usenix.org/supporters FALL 2020 VOL. 45, NO. 3 EDITOR EDITORIAL Rik Farrow 2 Musings Rik Farrow MANAGING EDITOR Michele Nelson SYSTEMS COPY EDITORS 6 An Empirical Guide to the Behavior and Use of Scalable Steve Gilmartin Persistent Memory Jian Yang, Juno Kim, Morteza Hoseinzadeh, Amber Ankerholz Joseph Izraelevitz, and Steven Swanson PRODUCTION 12 How to Not Copy Files Yang Zhan, Alex Conway, Nirjhar Mukherjee, Arnold Gatilao Ian Groombridge, Martín Farach-Colton, Rob Johnson, Yizheng Jiao, Ann Heron Michael A.
    [Show full text]