Loading…
To Learn More or Register: LinuxCon North America | CloudOpen North America
View analytic

Sign up or log in to bookmark your favorites and sync them to your phone or calendar.

developer [clear filter]
Wednesday, August 20
 

11:15am

Kernel Features for Reducing Power Consumption on Embedded Devices - Krzysztof Kozłowski, Samsung
Embedded devices must accommodate opposite requirements: high performance, low latency and low power consumption. Krzysztof Kozlowski will describe features of Linux kernel allowing reducing the power consumption on mobile devices (smartphones, tablets, wearables) without hurting the performance or latency. He will show what kernel configuration options are important, in the field of reduced power consumption, when preparing the kernel for ARM embedded devices. Although the talk will focus on Samsung's Exynos System-on-Chip family, presented ideas can be reused on other ARM embedded devices. Krzysztof will also show how certain features of SoCs match with kernel power management.

Survey this Session   

Speakers
KK

Krzysztof Kozłowski

Linux Kernel Developer, Samsung Electronics
Krzysztof Kozlowski has been working in Samsung R&D Institute Poland since 2009. His current job is development of Linux kernel for Tizen mobile devices (including wearables).


Wednesday August 20, 2014 11:15am - 12:05pm
Sheraton Ballroom III

11:15am

Linux Performance Tools - Brendan Gregg, Netflix
There are many performance tools nowadays for Linux, but how do they all fit together, and when do we use them? This talk summarizes the three types of performance tools: observability, benchmarking, and tuning, providing a tour of what exists and why they exist. Advanced tools including those based on tracepoints, kprobes, and uprobes are also included: perf_events, ktap, SystemTap, LTTng, and sysdig. You'll gain a good understanding of the performance tools landscape, knowing what to reach for to get the most out of your systems.

Survey this Session   

Speakers
avatar for Brendan Gregg

Brendan Gregg

Senior Performance Architect, Netflix
Brendan Gregg is an industry expert in computing performance and cloud computing. He is a senior performance architect at Netflix, where he does performance design, evaluation, analysis, and tuning. He is the author of Systems Performance published by Prentice Hall, and received the... Read More →


Wednesday August 20, 2014 11:15am - 12:05pm
Sheraton Ballroom I

12:15pm

CoreOS: An Introduction - Brandon Philips, CoreOS
The architectural patterns of a large scale platform are changing. Dedicated VMs and configuration management tools are being replaced by containerization and new service management technologies like systemd. This presentation will be giving an overview of their key technologies, including etcd, fleet, and docker. Come and learn how to use these new technologies to build performant, reliable, large distributed systems.

Survey this Session  

Speakers
JP

Johan Philippine

CEO, CoreOs
Alex Polvi is the CEO of CoreOS, a Y-Combinator funded start-up, focusing on building a new operating system for massive server deployments. Prior to CoreOS Alex was GM for Rackspace Hosting, Bay Area, overseeing cloud product development. Alex joined Rackspace through the acquisition... Read More →


Wednesday August 20, 2014 12:15pm - 1:05pm
Arkansas

12:15pm

Coming Soon, an Open Source Project Near You – the Linaro LNG Open Data Plane Initiative - Michael Christofferson, Enea
ARM formed the Linaro consortium in 2010, a non-profit organization composed now of over 20 industry leading companies. Linaro's aim is to provide an open source common software and tools foundation for the industry to use - making it easier and quicker to develop Linux-based products. In 2013 Linaro formed a “sub-group” called the Linaro Networking Group (LNG). LNG with twelve industry leading hardware and software vendors has since launched a new initiative called “Open Data Plane”, or ODP. ODP is an initiative designed to promote an open API specification, and an open source and cross platform implementation for high performance data plane packet processing that is independent of current solutions that require software or hardware vendor “lock-in”, like Intel's DPDK and Freescale's USDPAA, etc. This presentation is an overview of a) what ODP is, and b) the current status of ODP.

Survey this Session   

Speakers
avatar for Michael Christofferson, Enea

Michael Christofferson, Enea

Product Marketing Director, Enea
Previous speaking: 2014 - 1) ELC San Jose, April 29-May 2, 2014, "User Space Drivers in Linux – Pros, Cons, and Implementation Issues. In 2013 - 1) Linux Embedded Systems Conference San Franciso, Feb 20-22 2013 on the "Yocto Meta-Virtualization Project", 2) LinuxCon/Cloud Open North... Read More →


Wednesday August 20, 2014 12:15pm - 1:05pm
Parlor C

12:15pm

FireFoxOS and the Linux Kernel - Aimee Maree Forsstrom
FireFoxOS has hit the maket but what makes it different from other mobile phone platforms and how is it using the Linux Kernel? This talk aims to explain the architecture with a strong focus on GONK (Linux Kernel) that is the HAL component of the technology stack for the platform. This is a technical talk and will aim to dive deep into how information etc is passed between layers, Bootstrapping, init and userspace processes. I will also breifly cover how to contribute to the project and where our code bases live.

Survey this Session   

Speakers
AM

Aimee Maree Forsstrom

Technical Lead and Solutions Design, NSW State Library Australia
Aimee has worked in the IT Industry since 200 across the areas of Networking Engineering and Software Development. A true believe in Open Source and the benefits to society, people and business she was an early Open Source advocate building new Code management practices for Government... Read More →


Wednesday August 20, 2014 12:15pm - 1:05pm
Sheraton Ballroom III

12:15pm

Make your Own USB Gadget - Andrzej Pietrasiewicz, Samsung
A USB gadget can be connected to a host to extend it with additional functions. Creating a gadget means deciding what configurations there are and which functions each configuration provides. So far the choice has been made at kernel compile time, now it can be made with configfs instead at runtime. Composing a gadget of existing functions does not involve USB maintainers any more, basic shell scripting is enough, which opens new interesting possibilities for developers. This presentation is for Linux-based USB gadgets' vendors, Linux distributors and kernel/software developers. It is about how to create gadgets with configfs, including support for "OS descriptors" expected by a popular operating system, examples are given, and about how configfs is integrated into the gadget framework. Status update since last year is given and userspace's response to the new interface is described.

Survey this Session   

Speakers
avatar for Andrzej Pietrasiewicz

Andrzej Pietrasiewicz

Developer, Samsung R&D Institute Poland
Andrzej Pietrasiewicz graduated from Warsaw University of Technology, Faculty of Electronics and Information Technology, Warsaw, Poland in 2002. From then on he had been developing systems in C++ for over 5 years. Then for 3 years, he had been involved in various smaller projects... Read More →


Wednesday August 20, 2014 12:15pm - 1:05pm
Sheraton Ballroom I

2:30pm

IPMI - Automatically Configuring (and Compromising) Large Server Deployments - Matthew Garrett, Nebula
The Intelligent Platform Management Interface (IPMI) is a specification allowing system managers to control and query system state without relying on the host operating system. At its most basic, this allows for remote power management of servers - at its most advanced, it provides full control over firmware configuration, sensor output and remote access to the installed OS.

This presentation will cover the advanced features provided by many server vendors and introduce a new vendor-neutral interface for configuration of systems, making building clouds easier than ever. It will also cover security considerations relevant to IPMI and describe the risks associated with any failure to design deployments with sufficient paranoia.

Survey this Session   

Speakers
MG

Matthew Garrett

Staff Security Developer, Google
Matthew Garrett is a security developer at Google, working on infrastructural security for Linux desktop and mobile platforms.


Wednesday August 20, 2014 2:30pm - 3:20pm
Mississippi

2:30pm

Large Scale Identification of Race Conditions (In OpenStack CI) - Joe Gordon, HP
Does your project have a CI system that suffers from an ever-growing set of race conditions? We have the tool for you: it has enabled increased velocity despite project growth.
Richard Stallman once said, “it turned out that debugging these asynchronous multithreaded programs was really hard.” With 30+ asynchronous services developed by over 1000 people the OpenStack project is an object lesson of this problem. One of the consequences is race conditions often leak into code with no obvious defect. To address this problem we have developed an ElasticSearch based toolchain called “elastic-recheck.” This helps us track race conditions so developers can fix them and identify when CI failures are related to the failed patch or are due to a known pre-existing race condition. elastic-recheck has saved countless hours while allowing us to better manage our race conditions.

Survey this Session   

Speakers
avatar for Joe Gordon

Joe Gordon

Pinterest
Joe Gordon is an SRE at Pinterest, where he works on search and performance.. Before that He spent the last 4 years working full time on the open source project, OpenStack. Where he focused on improving quality. He has spoken at, and co-chaired at OpenStack summits. And has given... Read More →


Wednesday August 20, 2014 2:30pm - 3:20pm
Colorado

2:30pm

Replicating from RDBMS into Hadoop - MC Brown, Continuent
Big Data and Hadoop in particular is currently big business, but one of the most significant problems is getting your data into Hadoop from your existing RDBMS stores. Whether you have user profile information, trades, or transactional information from your webstore in your RDBMS, analysing it often takes place within Hadoop. Dump and load techniques imply delays and intermittent replication. With Tungsten Replicator we can move data from multiple MySQL and Oracle databases directly into Hadoop giving you carbon-copies of your RDBMS data in real-time. In this presentation I'll demonstrate how the replicator achieves this including a live demo, and describe how it has helped Groupon and Booking with the data migration needs, and where we are headed in the future for supporting different tools and techniques for replication.

Survey this Session   

Speakers
MB

MC Brown

Senior Information Architect, Continuent
MC Brown is a technology user and writer who has written over 26 books, contributed to many others, and written countless articles on topics as diverse as Microsoft, Linux (including time as LAMP Technologies editor for LinuxWorld magazine), Mac OS, Grid systems and Big Data, databases... Read More →


Wednesday August 20, 2014 2:30pm - 3:20pm
Sheraton Ballroom IV

2:30pm

RAS Enhancement Activities for Mission-critical Linux Systems - Yoshihiro Yunomae, Hitachi
Yoshihiro Yunomae will share bug fixing and suggestion of new features on the current upstream kernel for mission-critical systems such as banking systems and train management systems. People expect stable operation for long-term use, so the systems need to satisfy following RAS requirements:
1. To prevent any problems beforehand.
2. To continue the operation even if a problem is occurred.
3. To find out the root cause of the problem certainly in order to never induce the problem again.
He investigated the current upstream kernel from the viewpoints of these requirements. Then, he has found some bugs and lack of some features. For examples, there were deadlocking problem after panic() to get memory dump, linebreaks in a single error message in SCSI layer and so on. In this presentation, possible problems and the way to avoid the problems are to be explained in a detail way.

Survey this Session   

Speakers
YY

Yoshihiro Yunomae

Software Engineer, Hitachi Ltd.
Yoshihiro Yunomae is a Software Engineer at Hitachi Ltd. since 2010, he develops highly reliable Linux for mission-critical systems.


Wednesday August 20, 2014 2:30pm - 3:20pm
Sheraton Ballroom I

3:30pm

Cloud Operating Systems for Servers - Michael Day, IBM
Cloud Operating Systems for servers are optimized to deploy and run back-end infrastructure and middleware as efficiently as possible. Typically a Cloud OS will include just enough functionality in the host to run workloads in a Linux container or a KVM Virtual Machine.
Mike will discuss what makes a good Cloud Server OS and why Linux is providing the leading implementations, and is clearly providing innovation to the industry. He will also cover the two primary methods used by Cloud operating systems to reduce overhead: sharing a kernel using Linux Containers, or running on a bare-bones kernel such as OSv. (http://osv.io)
Mike will provide performance data for a range of “cloud operations” including deployment of workloads, performance measured by cpu, network, disk, and generally; using Docker with CoreOS and Atom, and OSv with KVM.

Survey this Session   

Speakers
MD

Michael Day

Distinguished Engineer and Virtualization Architect, IBM
Mike Day is an IBM Distinguished Engineer and is the Virtualization Architect for IBM's Open Systems Group. Mike has developed network operating systems, internet protocols, systems management software and hardware, security protocols, and virtualization software. Mike lead IBM's... Read More →


Wednesday August 20, 2014 3:30pm - 4:20pm
Colorado

3:30pm

Kimchi: An On Ramp KVM Management Tool - Zhengsheng Zhou, Shaohe Feng, Royce Lv, IBM
As we all know the KVM is good performance, and we have many kvm based virtualization management tools, openstack, ovirt, virt-manager, virsh. None of them is design for people with little knowledge of virtualization/linux. They need to equipped with knowledge of storage, network, etc, and with many efforts of enviroment setup--database, storage and so on.

We do need an on-ramp kvm management tool to make a fresh user to get started with KVM and create his first guest as easy as possible, and to attract more and more fresh users.
Kimchi is also an HTML5 based management tool for KVM. It enables users to access VMs from various platforms including laptop, tablet, smart phone, and from various OSes including Windows,Linux and iOS.

we will introduce our self-discovery conception and netflow oriented network conception from user perspective.

Survey this Session  

Speakers
SF

Shaohe Feng

China, IBM
Shaohe is a software developer from IBM, working on the Kimchi project. Kimchi is a HTML5 based management tool for KVM.


Wednesday August 20, 2014 3:30pm - 4:20pm
Huron

3:30pm

Ever Growing CPU States: Context Switch with Less Memory and Better Performance - Fenghua Yu, Intel
Context switch is a key part of multiprocessing. With more CPU features implemented, the switched states are growing. On x86 CPU, extended state (xstate) manages the state components which have been expanding from legacy floating points and SSE to AVX, AVX2, AVX512, MPX, and Intel PT registers. But the increasing xstates are putting pressure on mem/cache footprint for multi processes and speed of switching processes.
Fenghua Yu will overview existing context switch policy and its problems. He will discuss new technology to occupy less context mem and improve context switch performance. His optimization can reduce xstate mem space (2.5KB/process on the latest CPU) by 77% in the best case. New APIs to save/restore less context improve performance by limiting switched data to reduce time and cache footprint. The APIs are used in crypto, fast copy pages, etc.
The patches will be in 3.16.

Survey this Session  

Speakers
avatar for Fenghua Yu

Fenghua Yu

Linux Kernel Developer, Intel
Fenghua Yu is a Linux kernel developer in Intel. He has been working on Linux kernel development for over ten years. His projects cover various areas include platform QoS, EFI, kernel optimization, power management, context switch, security, etc. He lives in the San Francisco bay... Read More →


Wednesday August 20, 2014 3:30pm - 4:20pm
Sheraton Ballroom I

3:30pm

Kernel Debugging and Tracing - Jason Wessel, Wind River Systems
Jason will provide a state of the union with the kernel debugger (KGDB/KDB). And open the discussion of what people want to see with the future of debugging and kernel tracing. What are your kernel debugging pain points?

Survey this Session   

Speakers
JW

Jason Wessel

Product Architect, Wind River
Jason is the product architect for Wind River Linux and has been building embedded software solutions for the last 14 years. Going back more than 2 decades, he has worked on a wide range of Linux projects spanning from VR systems, web servers, proxies, build systems, and qemu, to... Read More →


Wednesday August 20, 2014 3:30pm - 4:20pm
Arkansas
 
Thursday, August 21
 

10:45am

Security in the Cloud: Containers, KVM, and Xen - George Dunlap, Citrix Systems UK Ltd
In our interconnected world of mobile and cloud computing, particularly with the rise of governmental spying, corporate espionage, and theft of data by organized crime syndicates, security is more important than ever. Many claims are being made about the security of open-source cloud technologies: How can administrators, users, and developers separate fact from fiction?

This talk will equip the audience with the principles needed to evaluate security claims. We will talk the nature of risk, of vulnerabilities and exploits; the various factors that reduce the risk of vulnerabilities in software; and about TCB, threat models, and defense-in-depth.

We will then apply these principles to three open-source cloud technologies: containers, KVM, and Xen, to see how they stack up. These will be backed up with numbers: lines of code, security advisories, entry points, and so on.

Survey this Session   

Speakers
avatar for George Dunlap

George Dunlap

Principle Software Engineer, Citrix Systems R&D UK Ltd
George Dunlap worked with the Xen project while a graduate student at | the University of Michigan before receiving his PhD in 2006. He is | currently working as Staff Software Engineer for Citrix on the | open-source Xen team in Cambridge, England. He has done work in many... Read More →


Thursday August 21, 2014 10:45am - 11:35am
Mayfair

10:45am

The Evolution of Open vSwitch - Jesse Gross, VMware
Open vSwitch is the primary open source virtual switch used in next-generation datacenter deployments. It runs on both software and hardware switching platforms and has seen adoption by a wide variety of emerging technologies such as OpenFlow, OpenStack, and different forms of network virtualization.

As Open vSwitch has moved from a small project to production infrastructure, the expectations and requirements placed on it have radically evolved. The result is a balancing act - how can it support the stability and performance requirements of a data center, while still maintaining openness and flexibility to enable new technologies?

Starting with the original design goals of Open vSwitch, this talk will give some background on where we've been, the challenges and progress that's been made, and future directions.

Survey this Session   

Speakers
JG

Jesse Gross

Sr. Staff Engineer, VMware
Jesse Gross has worked on the Open vSwitch project since its inception and is the Linux kernel maintainer of the fast-path dataplane. He is also a coauthor of several other technologies related to network virtualization including the Geneve tunneling protocol currently being standardized... Read More →


Thursday August 21, 2014 10:45am - 11:35am
Huron

10:45am

Kpatch Without Stop Machine - Masami Hiramatsu, Hitachi Ltd.
Kpatch is a kernel livepatching mechanism, which can replace kernel function on-line. Current kpatch uses stop_machine to ensure no thread is running on the old function and to atomic switching.
However, stop_machine has a scalability issue that the machine stop time will be bigger in multi-core system. Thus the stop_machine-less kpatch is required especially for such big-machine enterprise users. To solve this issue, this presentation shows the kretprobe-based function-user checker in kpatch, and upstream kprobe/kretprobe improvements for that purpose.

Survey this Session   

Speakers
MH

Masami Hiramatsu

Researcher, Hitachi Ltd.
Masami Hiramatsu is a Japanese kernel maintainer of kprobes/ftrace/perf-probe etc. He is working for Hitachi Ltd. and is a researcher in Yokohama Research Laboratory. He started working on Linux kernel with Kernel Tracing (LKST) at 2002, and joined to SystemTap development and became... Read More →


Thursday August 21, 2014 10:45am - 11:35am
Arkansas

10:45am

Object-Based Storage - NAS Support Through NFS-Ganesha. - Venkateswararao Jujjuri, IBM
Data growth is exploding and companies seeing a 60 to 70 percent of data growth every year. This trend likely to continue through the foreseeable future. Most of this data is unstructured data which is massive in scope and growing at a significantly faster rate than structured data. file-based storage method may not scale quickly to meet this demand and that is where object based storage comes in.

NFSv4.1 introduced pNFS which provides parallel data access to multiple classes of data including file-based and object-based storage. As the industry is transitioning, having the flexibility to serve multiple storage classes is a huge advantage. NFS-Ganesha, a user-level NFS server is working towards supporting access to different data classes and making NAS ready for next wave of storage revolution.

In this presentation I plan to introduce NFS-Ganesha and how it is serving Object store

Survey this Session   

Speakers
avatar for Venkateswararao Jujjuri

Venkateswararao Jujjuri

Cloud Storage Architect, Salesforce
Venkateswararao Jujjuri (JV), Architect at Salesforce and Apache BookKeeper committer, is responsible for building cloud storage services. He is currently working with Apache BookKeeper and Ceph Opensource projects. Prior to this, at IBM’s Linux Technology Center, worked on NFS-Ganesha... Read More →


Thursday August 21, 2014 10:45am - 11:35am
Sheraton Ballroom I

11:45am

OpenStack Horizon: Controlling the Cloud Using Django - David Lapsley, Metacloud
The OpenStack project was launched by Rackspace and NASA in July 2010. Since then it has gained considerable momentum, with over 200 companies joining the project.

Horizon is OpenStack’s web-based user interface. It is constructed in two parts: 1) a core set of libraries for implementing a dashboard; 2) a reference dashboard implementation that uses the core set of libraries. Customization is key to the Horizon framework. It allows developers to construct their own dashboards, panel groups, and panels, and assemble them via a common navigation/presentation framework.

In this presentation, David will provide a brief introduction to OpenStack and Horizon. Then he will review Horizon’s architecture, explain how it integrates with other OpenStack services, examine its most interesting features, and describe how to start developing with it.

Survey this Session  

Speakers
DL

David Lapsley

Director of Engineering, Metacloud
David Lapsley is the Director of Engineering at Metacloud. David has been using Python for over 10 years and Django for 5 years. His first commercial Python project was writing a Spread Spectrum Software Defined Radio application for NASA. His first commercial Django project was the... Read More →


Thursday August 21, 2014 11:45am - 12:35pm
Mississippi

11:45am

Better Integration of Systems Management Hardware with Linux - Charles Rose, Dell
Managing server platforms in data centers can be complex with the myriad hardware vendors and the variety of management consoles that manage them. Anything that reduces complexity would be welcome. Specifications/standards help, but there might still be a need for software specific to each server vendor running on the OS to ensure tighter integration with their respective systems management hardware. This adds to the complexity in the number of software components that need to be deployed/managed across multiple server vendors. Open source alternatives exist.
In this presentation, Charles Rose will review features in open source components that he has been involved with that enhance the OS - systems management hardware integration. This results in an ecosystem that relies on open source tools readily available in popular Linux distribution.

Survey this Session   

Speakers
CR

Charles Rose

Engineer, Dell Inc.
Charles Rose is an Engineer working in Dell's Operating Systems Engineering group. He has been working/associated with Linux since 1998. His responsibilities at Dell include ensuring the best experience for Enterprise customers using various flavors of Linux on Dell PowerEdge Server... Read More →


Thursday August 21, 2014 11:45am - 12:35pm
Sheraton Ballroom I

2:30pm

Fleet: systemd for Clusters - Brandon Philips, CoreOS
Scaling an application from one machine to two is a difficult. But, with it comes the benefits of fault tolerance and the opportunity to less expensive hardware. To get there you need the right tools.

fleet helps you scale work across a group of machines while preserving an interface all developers and sysadmins are familiar with: the init system. Using a simple command line tool you can schedule and run containers across a group of machines and make sure those processes continue to run somewhere in the cluster in the face of machines going down.

Learn how fleet can help you manage your machines and the technologies that make it possible.

Survey this Session  

Speakers
JP

Johan Philippine

CEO, CoreOs
Alex Polvi is the CEO of CoreOS, a Y-Combinator funded start-up, focusing on building a new operating system for massive server deployments. Prior to CoreOS Alex was GM for Rackspace Hosting, Bay Area, overseeing cloud product development. Alex joined Rackspace through the acquisition... Read More →


Thursday August 21, 2014 2:30pm - 3:20pm
Mississippi

2:30pm

Open Source Policy: OpenDaylight and OpFlex - Scott Mann, Cisco
Policy is an increasingly important area of research and development. There is a growing interest in using policy-based systems for IT automation. Witness DevOps tools such as Puppet, Chef, CFengine, and Ansible. Software Defined Networking is another area with policy based R&D taking place. Much like DevOps and Software Defined Networking, policy will be driven by Open Source platforms. In this talk we will discuss OpFlex, an open distributed control system based on a declarative policy information model. OpFlex is an incubated project within the OpenDaylight project under active development. This talk will cover the OpFlex project in OpenDaylight, current status and future direction, along with an overview of how OpFlex can integrate with an IaaS platform such as OpenStack.

Survey this Session   

Speakers
SM

Scott Mann

Cisco
Scott works at Cisco Systems on Open Source projects ranging from OpenStack to Open vSwitch and OpenDaylight. His recent work is around policy in the Open Source domain. He contributes to OpFlex, which is an incubated project in OpenDaylight. He is working to drive an open, scalable... Read More →


Thursday August 21, 2014 2:30pm - 3:20pm
Arkansas

2:30pm

Open Innovation with IBM and The OpenPower Foundation's Coherent Accelerator Processor Interface - Brian Zoric & JT Kellington, IBM
Brian Zoric and JT Kellington from IBM  will be presenting the knowledge required to create innovative computing solutions based around the POWER8 I/O protocol known as  "Coherent Accelerator Processor Interface" (or CAPI for short).  A CAPI connected device hooked up to a POWER8 microprocessor creates a "hollow core", or programmable processor, that can dramatically accelerate any data-intensive workload you can imagine. This technology opens the door to innovation from the industry and academia in a profound manner. We have barely scratched the surface of what is possible with this new technology. Specific CAPI coding examples will be discussed showing its ease of use, and attendees will exit the talk equipped with the knowledge that they need to immediately start  innovating with this new technology. 

This talk will outline what CAPI is , some ways it has been used in the lab and will cover some specific scenarios and proof points showing the end user benefits from this exciting new acceleration technology.  In addition, specific CAPI coding examples will be discussed showing the  ease of use of the technology. Attendees will exit the talk equipped with the knowledge that they need to immediately start innovating with this new technology.

Speakers
JJ

Jeffrey (JT) Kellington

JT Kellington is currently the Coherent Accelerator Processor Interface (CAPI) lab lead for the IBM POWER Systems design team. He is also actively contributing to the development and debug of the Linux kernel stack for CAPI enablement in POWER systems. He graduated from the University... Read More →
BZ

Brian Zoric

Brian Zoric currently manages the executive briefing centers in both Austin, TX and Rochester, MN. Previously Brian managed the Chief Architects and Chief Engineers of all POWER Systems. Prior career roles have included chip integration, circuit design and computer aided design... Read More →


Thursday August 21, 2014 2:30pm - 3:20pm
Sheraton Ballroom V

3:30pm

An Overview of Kernel Lock Improvements - Davidlohr Bueso, & Scott Norton, HP
Synchronization primitives (locks) are at the heart of any system, and the Linux kernel is no exception. While lockless algorithms/data structures are an attractive alternative with performance in mind, reality dictates otherwise, many times due to the extra complexity involved. In the kernel, lockless structures are, rather the exception, than the norm.

Davidlohr Bueso will review recent changes and discussions on the topic, including mutexes, rwsems and futexes. The algorithmic ideas behind them will be revisited and uncover some of the issues that can significantly impact overall system scalability. This work has led to important performance benefits on large multi-core systems, spanning a variety of workloads and configurations. Future room for improvements and a look forward on how to address them will also be covered.

Survey this Session   

Speakers
DB

Davidlohr Bueso

Software Engineer, SUSE Labs
SN

Scott Norton

Scott Norton is a Master Technologist at Hewlett-Packard with a focus on the performance and scalability of the Linux kernel for large x86 NUMA based HP servers. He has been a kernel engineer at HP for the Linux, HP-UX, and NonStop kernels since 1988 with a focus on process/thread... Read More →


Thursday August 21, 2014 3:30pm - 4:20pm
Arkansas

3:30pm

Hacking OpenStack for Padawans - Adolfo Brandes, Hastexo
This is where you'll learn how to hack OpenStack itself. You'll start by firing up your own OpenStack development environment, go on to modify OpenStack itself, and end the session learning how to contribute your modification back to the project. Sounds like a lot? That's because it is!

Part I. Setup: You'll get a set of ready-to-use VirtualBox appliances as well as instructions on how to set them up on your machine. At the end of this initial part of the session you should have everything you need to start hacking!

Part II. Hacking: During this section you'll see how to set up an environment for comfortable OpenStack hacking, modify a small part of OpenStack, and finally run and test the modification.

Part III. Contribution: Armed with your patch, during this section you'll step through the process of submitting it to the OpenStack community for review.

Survey this Session   

Speakers
avatar for Adolfo R. Brandes

Adolfo R. Brandes

Hastexo
Adolfo Brandes will be your trainer. Adolfo started out as an English teacher, but a talent for programming computers shaped his early career. He developed everything from an Asterisk-based PBX to the odd Linux kernel driver, including several websites and an AGPL3 Python-backed web... Read More →


Thursday August 21, 2014 3:30pm - 5:20pm
Mississippi

4:30pm

Atomic in Depth: A Look at Project Atomic for Managing Docker Containers - Joe 'Zonker' Brockmeier, Red Hat
Project Atomic, a concept for creating hosts to run Docker containers, was unveiled in April of this year. Rather than producing its own derivative distribution, Atomic is an umbrella project for creating hosts that are optimized for running Docker containers, with an "atomic" update model for the host operating system.

This talk will provide an overview of Atomic along with an in-depth tour of the features offered by an Atomic host, the tools used for creating and updating hosts, and managing Docker containers more effectively. We'll also discuss the implementations of Atomic (Fedora, CentOS, and RHEL) and how to get involved in the Atomic community.

Survey this Session   

Speakers
avatar for Joe Brockmeier

Joe Brockmeier

Red Hat, Red Hat
Joe Brockmeier is a long-time participant in open source projects and former technology journalist. Brockmeier has worked as the openSUSE Community Manager, is an Apache Software Foundation (ASF) member, and participates heavily in the Fedora Cloud Working Group. Brockmeier works... Read More →


Thursday August 21, 2014 4:30pm - 5:20pm
Sheraton Ballroom III

4:30pm

Compressing Strings of the Kernel - Wolfram Sang
In the past, there have been experiments to compress the strings within the kernel to reduce run-time memory footprint. However, not much more than concepts and estimations have been done, yet those results were interesting enough to be further researched. This presentation shows the results of my research. It will include topics like how to obtain the strings, comparison of compression techniques, and implementation details under the light of upstream acceptance. Furthermore, results of the data analysis will provide guidelines for developers how to add strings and other kinds of data, in order to reduce memory footprint and bloat, in general.

Survey this Session   

Speakers
WS

Wolfram Sang

Upstream Kernel Developer & Maintainer, Renesas / Consultant
Wolfram Sang has been working as a Linux kernel developer for embedded systems since 2008. He maintains the I2C subsystem and works as a consultant, mainly for the Renesas Upstream Kernel Team. Programming since his childhood, he still hacks his machines from the 80s, especially the... Read More →


Thursday August 21, 2014 4:30pm - 5:20pm
Sheraton Ballroom V

4:30pm

Getting More Out Of System Suspend In Linux - Rafael J. Wysocki, Intel OTC
System suspend is one of the most mature and widely used power management techniques in Linux. At the same time the kernel's system suspend infrastructure is susceptible to inadvertent errors, so major changes in that code are relatively rare and applied with caution. However, that conservative approach has caused system suspend to gradually fall behind with meeting the expectations of modern systems' users. It has turned out recently that it needs to be faster, integrated more tightly with runtime PM, more careful about avoiding unnecessary operations and capable of handling systems with more advanced PM support in hardware in better ways. As a result, several efforts are under way to improve system suspend in Linux and make it even more useful. I will discuss those efforts, describe the improvements that have been achieved already and outline possible directions of future development.

Survey this Session   

Speakers
avatar for Rafael J. Wysocki

Rafael J. Wysocki

Software Engineer, Intel OTC
Rafael maintains the Linux kernel's core ACPI and power management code, including the core infrastructure for IO device PM, CPU PM and system suspend/hibernation. He works at Intel Open Source Technology Center as a Software Engineer focusing on the mainline Linux kernel. Rafael... Read More →


Thursday August 21, 2014 4:30pm - 5:20pm
Arkansas

4:30pm

SteamOS Magic - Keith Packard, Intel
Turning the Linux Desktop into a Gaming Console Valve Software, developers of the Steam app-store for desktop computer games, started offering their games and the Steam environment for Linux users back in February of 2013. Since then, they've gone further and constructed a custom Debian-derivative Linux distribution that turns a regular desktop PC into a gaming console, similar in concept to the original Microsoft Xbox, which was essentially a standard PC running a modified version of Windows. However, unlike the Xbox, SteamOS is designed to run unmodified versions of existing Linux games offered through the Steam store. This presented two separate challenges, the first to provide a game console experience starting with the standard X Window System, and the second to take desktop games and transform them.

Survey this Session   

Speakers
KP

Keith Packard

Distinguished Linux Technologist, HP
Keith Packard has been developing free software since 1986, working on | the X Window System, the Linux kernel and rocketry electronics. He is | currently a Distinguished Linux Technologist at HP working as the | Chief Architect for Linux on The Machine. Keith received a Usenix... Read More →


Thursday August 21, 2014 4:30pm - 5:20pm
Sheraton Ballroom I
 
Friday, August 22
 

10:45am

Beating the Virtualization Tax for NoSQL Workloads With OSv - Don Marti & Glauber Costa, Cloudius
NoSQL data stores such as Cassandra and Redis are memory and CPU intensive loads which can have unacceptable performance when run under virtualization in production environments.

Over the past two years, the OSv project has profiled some of the important bottlenecks for NoSQL applications, and designed a new open-source guest environment to improve performance in key areas such as JVM memory allocation and network throughput.

The side effect of this performance work has been to move some difficult tuning tasks off the administrator's to-do list and into the guest OS where they can be handled automatically with better responsiveness.

Survey this Session   

Speakers
GC

Glauber Costa

Lead Software Engineer, Cloudius System
Glauber Costa is a Lead Software Engineer at ScyllaDB, where he helps developing the Scylla NoSQL database and the Seastar framework. He was also one of the early engineers responsible for OSv, and has extensive past experience in the Linux Kernel containers implementation and Open... Read More →
DM

Don Marti

Technical Marketing Manager, Cloudius Systems
Don Marti is a technical marketing manager for Cloudius Systems, the OSv company. He has written for Linux Weekly News, Linux Journal, and other publications. He co-founded the Linux consulting firm Electric Lichen, which was acquired by VA Linux Systems. Don has served as president... Read More →


Friday August 22, 2014 10:45am - 11:35am
Colorado

10:45am

Storage Tiering and Erasure Coding in Ceph - Sam Just, Red Hat
Ceph is designed around the assumption that all components of the system (disks, hosts, networks) can fail, and has traditionally leveraged replication to provide data durability and reliability. The CRUSH placement algorithm is used to allow failure domains to be defined across hosts, racks, rows, or datacenters, depending on the deployment scale and requirements. Ceph has introduced a storage tiering infrastructure and cache pools that allow alternate hardware backends (like high-end flash) to be leveraged for active data sets while cold data is transparently migrated to slower backends. The combination of these two features enables a surprisingly broad range of new applications and deployment configurations. This talk will cover a few Ceph fundamentals, discuss the new tiering and erasure coding features, and then discuss a variety of ways that the new capabilities can be leveraged.

Survey this Session   

Speakers
SJ

Sam Just

Developer, Red Hat
Sam is a Ceph contributor working for Red Hat in the Inktank group. Most of his work is focused on RADOS, the object store underneath Ceph. In his spare time he likes cycling and gaming. He is also very bad at typing about | himself in the third person.


Friday August 22, 2014 10:45am - 11:35am
Mississippi

10:45am

New Views on Your Source Code History with 'git replace' - Christian Couder, SoftAtHome
Git has become the most popular version control system in the Open Source world, and more and more companies are also using it. The source code history when managed by Git is supposed to be immutable, because Git uses a content addressed database. The Git objects are indexed by their SHA-1 hash. When mistakes have been made, or to make some history based features more useful or more reliable, though, it can be interesting to transform the Git source code history. To do that it is a good idea to use "git replace".

"git replace" makes it possible to have different "views" on the history, instead of only one. It works by creating special references and using these special refs to dynamically use some git objects instead of the original ones.

Survey this Session   

Speakers
CC

Christian Couder

Software Engineer, Software Engineer
Christian Couder is a software engineer with 20 years of professional experience in software development, release, build and version control. He is a Git developer since June 2006. He has been working on many part of Git, especially "git bisect" and lately "git rebase". Since March... Read More →


Friday August 22, 2014 10:45am - 11:35am
Sheraton Ballroom II

11:45am

Virgil3D: Journey to a Virtio GPU - Dave Airlie, Red Hat
There has been a gap in the Linux/KVM/qemu desktop virtualisation space for many years compared to a number of competing offerings, the lack of a paravirtual 3D accelerator device. This project is an attempt by one man to fix this large gap. Based on the Mesa internal hardware abstraction layer (Gallium3D) this project creates a virtio based GPU that can handle modern OpenGL and Direct3D requirements. The 3D rendering streams are passed to a renderer process in the host and it uses the local OpenGL interface to do the rendering. The project also encompasses a simple unaccelerated non-3D virtio gpu for situations where there is no local 3D hardware present.

Survey this Session  

Speakers
DA

Dave Airlie

Enterprise Kernal Graphics Maintainer, Red Hat
Dave Airlie is the kernel graphics maintainer looking after various parts of the Linux kernel associated with video drivers. He works on X.org and Mesa projects as a major contributor to graphics driver development. He is currently employed by Red Hat in Brisbane as the enterprise... Read More →


Friday August 22, 2014 11:45am - 12:35pm
Huron

11:45am

Standard Specification to Open Source in Internet of Everything - Guy Martin, Samsung
This presentation talkes about the introduction of Internet of Everything including a new open source community and our experience on Internet of Everything using standard specification and reason for open source toward a success of Internet of Everything. In particular, this presentation illustrates the various benefits that Internet of Everything can ream from open source based on the lesson learned from standard specification and its limitation. Finally, this presentations includes Samsung efforts on Intetnet of Everything and provide some proof that Samsung is doing what they are preaching to the future of Internet of Everything.

Survey this Session   

Speakers
GM

Guy Martin

Senior Open Source Strategist, Samsung Open Source Group
Guy is the Senior Open Source Strategist at Samsung.


Friday August 22, 2014 11:45am - 12:35pm
Sheraton Ballroom III

1:45pm

Is SSH Really Secure? - Mark Lambiase, Fox Technologies
SSH is nearly ubiquitous in today’s enterprises, and is the predominant tool for managing Linux servers and the applications and data that they host.  Poor practices around the deployment and management of the SSH infrastructure has been blamed or associated with various breaches, and has gained considerable attention from management and administrators. A number of reports over that last two years have exposed the vulnerabilities in SSH; not in the protocol or software itself, but in the poor practices deploying, managing and maintaining control over it.

SSH must be managed properly, in a way that accounts for its various methods of authentication, and in a way that accounts for its various features. But, even properly managing SSH alone does not actually ‘secure’ an SSH deployment.  SSH relies on accounts, and they must be accounted for in a security plan,  along with what people are allowed to do once they connect.

In this presentation, we will discuss the numerous pitfalls and common mistakes related to SSH deployment; as well as best practices in properly deploying SSH across your diverse server environment.

Survey this Session   

Speakers
ML

Mark Lambiase

As Chief Technology Officer, Mark Lambiase is responsible for Product Management, Engineering and Customer Services at FoxT. Mark comes to FoxT with over 15 years of experience designing, building, implementing, and supporting a broad range of identity, authentication and other related... Read More →


Friday August 22, 2014 1:45pm - 2:35pm
Sheraton Ballroom III

1:45pm

The Architecture of Linux Traffic Control - Cong Wang, Twitter
Traffic Control is an important and complicated part of Linux kernel networking subsystem, especially when buffer bloat becomes more and more important today. It is not easy to understand its design and architecture for beginners. In this presentation, Cong will give a short introduction to Linux Traffic Control, an overview of its design and architecture, and discuss the potential improvements for Linux Traffic Control in the future.

Survey this Session   

Speakers
CW

Cong Wang

Twitter Inc.
Cong Wang joined Twitter Inc. in 2013. He started to work on Linux networking since he was in Red Hat and has contributed many patches to Linux kernel since he was in college. Besides Linux kernel, he also contributed many patches to iproute2, libnl and other open source projects... Read More →


Friday August 22, 2014 1:45pm - 2:35pm
Sheraton Ballroom II

2:45pm

novm: Hypervisor Rebooted - Adin Scannell
Virtualization isn't only used to consolidate legacy hardware anymore; it runs the world's largest shared compute infrastructures. What if we reinvent the hypervisor for the cloud? Strong isolation, performance, and spin-up time matter -- not compatibility.

Over recent years, organizations have dramatically changed how they deploy and manage services, with Linux and virtualization at the center of the shift (AWS, OpenStack, etc.). Despite this, many developers and operators have minimal insight into how a hypervisor works or innovation possible at the bottom of the software stack.

The audience can expect a brief overview of virtualization fundamentals, followed by technical details of novm and specific insights gained during development, some straight-forward and others advanced. I hope that most developers will come away with new knowledge and curiosity, while advanced developers may be interested in some of the specific design decisions or roadblocks encountered.  

Survey this Session  

Speakers
avatar for Adin Scannell

Adin Scannell

Google
Adin is software engineer at Google, where he works on the technical infrastructure team. Previously, he was co-founder & CTO at Gridcentric, a virtualization technology company. Adin is a technologist with experience in academia and industry, including two industrial research labs... Read More →


Friday August 22, 2014 2:45pm - 3:35pm
Sheraton Ballroom IV

2:45pm

Backporting the kernel with SmPL - Luis R. Rodriguez, SUSE
As we evolve as software developers we not only have to consider how to better evolve
software but also how to more efficiently backport it. The Coccinelle engine was
developed to help evolve writing software faster and more efficiently, but can it also be
used to help backport? This talk will cover how SmPL was embraced by the Linux
backports project to help it backport the Linux kernel more effectively, more automatically,
and to help the effort scale.

This talk is intended for Linux kernel developers and kernel tree maintainers. No
experience with SmPL is required.

Survey this Session   

Speakers
LR

luis Rodriguez

Luis started hacking on the kernel since 2.6.5 through the first 802.11g driver upstream on the kernel, prism54. Since then he's moved on to address regulatoryc onsiderations on Linux and then a slew of 802.11 driver updates. Luis also maintains the Linux kernel backports project... Read More →


Friday August 22, 2014 2:45pm - 3:35pm
Ohio

2:45pm

LLVMLinux: Embracing the Dragon - Behan Webster, Converse in Code Inc.
This session will provide an update on the status of the LLVMLinux project; a project which is cooperating with both the Linux kernel and LLVM communities to build the Linux kernel with Clang/LLVM. This talk will also cover new things in LLVM which make clang even more attractive to the kernel community. LLVM is an extensive compiler technology suite which is already commonplace from Android/Renderscript and OpenCL through to high performance computing clusters.

Survey this Session   

Speakers
avatar for Behan Webster

Behan Webster

Chief Engineer, Converse in Code Inc
Behan Webster is a Computer Engineer who has spent the past two decades in diverse tech industries such as telecom, datacom, optical, automotive, and the game industry writing code for a range of hardware from the very small to the very large. Throughout his career his work has always... Read More →


Friday August 22, 2014 2:45pm - 3:35pm
Sheraton Ballroom III

3:45pm

Building Linux Support for Digital TV - Mauro Carvalho Chehab, Samsung
This speech will cover the infrastructure that is being built in order to support Digital TV on Linux, including Kernel and userspace library support.

Survey this Session   

Speakers
avatar for Mauro Carvalho Chehab, Samsung

Mauro Carvalho Chehab, Samsung

Linux Kernel Media Maintainer, Samsung
Mauro is the upstream maintainer of the Linux kernel media and EDAC subsystems, and also a major contributor for the Reliability Availability and Serviceability (RAS) subsystems. Mauro also maintains Tizen on Yocto packages upstream. He works for the Samsung Open Source Group since... Read More →


Friday August 22, 2014 3:45pm - 4:35pm
Sheraton Ballroom III

3:45pm

NFV in the Embedded World: Yocto Project and OpenStack - Jason Wessel, Wind River Systems
While there are many ways to build an embedded NFV solution, there are very few easy ways to get started until now. Using the Yocto Project plus the meta-cloud-services, you can build a complete OpenStack controller and compute node using the latest upstream or stable OpenStack. This presentation will cover why you would consider this approach to building an NFV solution and how to get started. Jason will demonstrate the capability of the Yocto Project configuration and layers. Guest instances can be scaled up or scaled down to build exactly what is needed for specific NFV purpose.

Survey this Session  

Speakers
JW

Jason Wessel

Product Architect, Wind River
Jason is the product architect for Wind River Linux and has been building embedded software solutions for the last 14 years. Going back more than 2 decades, he has worked on a wide range of Linux projects spanning from VR systems, web servers, proxies, build systems, and qemu, to... Read More →


Friday August 22, 2014 3:45pm - 4:35pm
Sheraton Ballroom I