Loading…
Learn more about Open Source Summit Japan and register here

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

Intermediate [clear filter]
Wednesday, May 31
 

11:50

Anatomy of Azure App Service on Linux: Bringing Containers into Existing PaaS Platform - Michimune Kohno & Nazim Lala, Microsoft
Supporting Linux opened up a whole new PaaS experiences at the Azure App Service which only supported Windows before. Particularly applying containers including Docker allows more people to run open source based solutions on our platform and provide easier integration with their existing software assets. In this session Michimune Kohno and Nazim Lala will discuss the best practices to take advantage of this new service, explain its internal architecture so developers can fully leverage the platform and obtain the best performance from it. They will also share some key learnings they have faced when building our platform.

Speakers
MK

Michimune Kohno

Senior Software Engineer, Cloud + Enterprise, Microsoft
Michimune Kohno, Ph.D. is a software engineer at Microsoft. He is currently working on the App Service Linux feature which allows developers to easily create web sites on Linux VMs and deploy contents to them with git, docker, etc. He gave a talk about the internal architecture of... Read More →
avatar for Nazim Lala

Nazim Lala

Principal Software Engineer, Microsoft
Nazim is a Principal Software Engineer working for Microsoft and is the developer lead and architect for Azure App Service on Linux, which is a fully managed Platform as a Service (Paas) to operate and manage web workloads in Azure. We have adopted an existing platform and containarized... Read More →


Wednesday May 31, 2017 11:50 - 12:30
Room 1

11:50

Leading an Open Source Community at a large Enterprise - Jonas Rosland, {code} by Dell EMC
Creating an open source initiative at a large enterprise such as Dell EMC comes with both challenges and rewards. Making sure your community is engaged and your projects thrive takes time and effort. In this session, Jonas Rosland, Open Source Community Manager at {code} by Dell EMC shares experiences, failures, and gives a glimpse into how large enterprises can embrace and lead open source communities successfully.

Speakers
JR

Jonas Rosland

Open Source Community Manager, {code} by Dell EMC
Jonas Rosland is a community builder, open source advocate, blogger and speaker at many open source focused events. As Open Source Community Manager at {code} by Dell EMC, he is responsible for the growth and prosperity of the {code} Community.


Wednesday May 31, 2017 11:50 - 12:30
Room 6

14:00

Device-DAX: Towards Software Defined Memory - Dan Williams, Intel
The "software defined" buzzword, to a Linux kernel developer, implies infrastructure that allows an application to bypass the kernel's general purpose interfaces and policies to compose a solution tailored to a specific use case. As performance and feature-differentiated memory technologies, like high-bandwidth memory and persistent memory, continue to proliferate the kernel needs a mechanism to hand out dedicated and predictable access to applications purpose built to leverage these new memory types. The DAX mechanism in general, and the Device-DAX interface in particular, allow an application direct-access to memory ranges bypassing the page cache and filesystem. This presentation gives an overview of the Device-DAX capability and how it might be used as a building block for software defined memory management.

Speakers
DW

Dan Williams

Intel
Dan is a Linux kernel developer in Intel Open Source Technology Center. He primarily works on enabling platform storage technologies. Most recently he has been involved in persistent memory enabling as a maintainer of the Linux kernel libnvdimm sub-system. He led the Persistent Memory... Read More →


Wednesday May 31, 2017 14:00 - 14:40
Hall B-1

14:00

Doing “DropBox” the Cloud Native Way - Krishnan Parthasarathi, Minio & Jörg Schad, Mesosphere
Cloud Native architecture has slowly become the default way to build robust, scalable applications. How would you rebuild a large scale storage service such as Dropbox (please note Dropbox is just an example here and we have no plans of actually competing with Dropbox), using the Cloud Native architecture? In this presentation, Joerg and Nitish will discuss the Cloud Native architecture, its advantages, and then explain how to build a scalable, multi-tenant, Dropbox like cloud storage system using modern, containerized applications like Minio as the unstructured data/blob store, MongoDB for structured data, Redis for key value pairs etc, all orchestrated on DC/OS. The presentation will also include a live demo of the Dropbox equivalent product.

Speakers
KP

Krishnan Parthasarathi

Senior Software Engineer, Minio, Inc.
Krishnan Parthasarathi is an engineer at Minio working on S3-compatible object storage server and SDKs. His interests include distributed systems and functional programming specifically Haskell.
avatar for Jörg Schad

Jörg Schad

Head of Machine Learning, ArangoDB
Jörg Schad is Head of Machine Learning at ArangoDB. In a previous life, he worked on built machine learning pipelines in healthcare, distributed systems at Mesosphere, and in-memory databases, conducted research in the Hadoop and Cloud area. He’s a frequent speaker at meetups... Read More →



Wednesday May 31, 2017 14:00 - 14:40
Room 1

14:00

Introduction to the Open Service Broker API- Doug Davis, IBM
CloudFoundry has been a leader in the PaaS space for many year. Integral to CF's success is the ability for apps to easily and seamlessly integrate with 3rd party services. Applications can have a loose coupling to their dependent services and CF will manage all aspects of the service's lifecycle for them. This powerful model has not gone unnoticed by the broader Cloud Native community. As a result, the CFF has joined forces with other key players in the community to create the new Open Service Broker API project to broaden CF's Service Broker API such that it can be used by other platforms (such as Kubernetes). This project will not only bring this SB model to other platforms but will also enable a new level of interoperability for Service Providers. This talk will introduce the new Open Service Broker API project, explain its mission, its members and its future plans.

Speakers
avatar for Doug Davis

Doug Davis

OM Knative, IBM
Doug works in IBM's Hybrid Cloud division. He's been working on Cloud related technologies for many years and has worked on many of the most popular OSS projects, including OpenStack, CloudFoundry, Docker and Kubernetes. He's currently the OM for Knative, co-chairing the CNCF's Serverless... Read More →


Wednesday May 31, 2017 14:00 - 14:40
Room 6

14:00

Linux Kernel ABI Specification - Sasha Levin, Verizon Labs
The ABI, the layer that joins the kernel and userspace is quite a mess. Various different interfaces, lacking documentation, and constant changes make it hard for anyone who uses the kernel to know what they can expect from the kernel when their userspace application makes a request. The purpose of the ABI specification project is to fully document the ABI interface in both a human readable and a machine readable form; this will allow verification that both the userspace application and the kernel behave as agreed in the "contract". This would also allow for more research into subsets of the kernel's ABI, and how to limit certain functions of the kernel by either allowing or blocking parts of the ABI.

Speakers
SL

Sasha Levin

Kernel Hacker, Microsoft
Sasha is a contributor to stable trees, the maintainer of the 4.1 LTS tree, and has previously maintained 3.18 LTS. Sasha is also the maintainer of liblockdep, a userspace lockdep library. Sasha is currently employed by Microsoft where he helps make Linux run better on Windows. Previously... Read More →


Wednesday May 31, 2017 14:00 - 14:40
Private Dining

14:50

Enhance Fuego Test Efficiency by Applying Additional Software & Hardware Solutions - Khiem Nguyen, Renesas Design Vietnam
Applying Fuego (Jenkins-based test automation) into open-source software verification is widely adopted recently. To enhance Fuego efficiency, Renesas is applying additional software solutions (like LAVA, TestLink) and hardware solutions (like boards supporting USB/SDCard hotplug).
The presentation shows Renesas activities through the demonstration of our current system, and provide feedbacks to Fuego upstream.

Speakers
avatar for Khiem Nguyen

Khiem Nguyen

Senior Staff Engineer, Renesas Design Vietnam
Khiem Nguyen has worked on Embedded software development for 10 years. He's a BSP developer for Open-Source Automotive Infotainment System. His team develops and enhances the functionality and performance of some Linux features like Power Management, Thermal Management, Memory Management... Read More →



Wednesday May 31, 2017 14:50 - 15:30
Private Dining

14:50

Noah: Hypervisor-Based Darwin Subsystem for Linux - Takaya Saeki & Yuichi Nishiwaki, The University of Tokyo
In this talk, we present Noah, a Linux binary compatibility layer for macOS based on hypervisor technology. Noah executes Linux ELF binaries by loading it into a hypervisor in which no kernel code is running. System calls invoked by the Linux program are trapped by the hypervisor and handled by the user program running on the host OS. Noah is designed to be more efficient, secure, and portable, compared to other Linux binary compatibility layers including Windows Subsystem for Linux (WSL). We will describe in detail how our hypervisor-based approach works and how linux system calls are translated. Performance comparisons with other related technologies will be presented for various practical situations. We will also discuss possible business applications of our compatibility technology.

Speakers
avatar for Yuichi Nishiwaki

Yuichi Nishiwaki

Yuichi Nishiwaki is a Master's student in Computer Science at the University of Tokyo. His research focuses on the theoretical aspects of computer systems. His recent work has been on distributed systems, programming languages, and kernel development. He is a creator of Noah, an... Read More →
avatar for Takaya Saeki

Takaya Saeki

Master's Student, The University of Tokyo
Takaya Saeki is a Master's student in Computer Science at the University of Tokyo in Japan. His research focuses on formal engineering methods. He loves all levels of computer programming, from low to high, from the kernel to the Web. He is a creator of Noah, a tool that makes unmodified... Read More →



Wednesday May 31, 2017 14:50 - 15:30
Hall B-1

16:00

Cloud Foundry GrootFS: A Daemonless Container Image Manager that You Can Use with RunC for Fun and Profit - George Lestaris, Pivotal
runC, OCI’s implementation of the runtime spec, has been very successful since the initiative was launched back in 2015. Cloud Foundry and others have been running with runC in production for a while now with hundreds of thousands of runC containers being spawned around the world on every day. For us, in Open Source conference, runC is quickly becoming the standard implementation of containers in Linux. runC, however, does not deal with the container packaging techniques which currently get standardised as well through the image spec initiative.

In this talk, George will introduce you to GrootFS, a deamonless container image manager, which can run seamlessly as a non-root user in Linux and provide runC with various types of container images. GrootFS is dealing with the low-level filesystem operations required to make container images efficient as well as the distribution of images.

Speakers
avatar for George Lestaris

George Lestaris

Product Manager,  Pivotal



Wednesday May 31, 2017 16:00 - 16:40
Room 1

16:00

Libral: Towards a Systems Management API for Linux - David Lutterkort, Puppet, Inc.
Linux is famous for not having a comprehensive management API. Over the last decade of implementing configuration management tools, we have learnt what such an API should look like, and produced multiple implementations of such API's. Yet, these implementations are only useful in the context of one specific tool. Past attempts at such an API have generally failed because they were too hard to use and their management capabilities too hard to extend. Libral aims to provide a management API that existing and future management tools can build onto. Its goals are a desired-state, idempotent management API, a footprint that makes it suitable for resource-constrained environments, and extension mechanisms that make it easy to add management of new types of resources. The API is bidirectional so that libral can be used both for making changes and retrieving current configuration.

Speakers
DL

David Lutterkort

Advisory Software Engineer, Puppet, Inc.
David is a software engineer at Puppet, where he’s worked on projects such as application orchestration and Razor, the best provisioning tool, ever. Before joining Puppet, David worked at Red Hat on a variety of management tools and served as the maintainer of Apache Deltacloud... Read More →


Wednesday May 31, 2017 16:00 - 16:40
Room 6

16:00

Understanding SCHED_DEADLINE - Steven Rostedt, VMware
Starting in Linux version 3.14, a new scheduling class was introduced. This class is called SCHED_DEADLINE. It implements Earliest Deadline First (EDF) along with a Constant Bandwidth Scheduler (CBS) that is used to give applications a guaranteed amount of CPU for a periodic period. This type of scheduling is advantageous for robotics, media players and recorders, as well as virtual machine guest management. This talk will explain the history of SCHED_DEADLINE and compare it with various other methods to deal with periodic deadlines. It will also discuss some of the current issues with the current Linux implementation and some of the improvements that are currently being worked on.

Speakers
avatar for Steven Rostedt

Steven Rostedt

Open Source Engineer, VMware Inc
Steven has been working on the Linux kernel since 1998 (started while working on his masters). He has been working on the Linux kernel professionally since 2001. Steven is one of the original developers of the PREEMPT_RT patch which turns Linux into a true real-time operating system... Read More →


Wednesday May 31, 2017 16:00 - 16:40
Hall B-1

16:50

Container Networking for Micro-services - An Apache Mesos Networking Deep Dive - Jörg Schad & Avinash Sridharan, Mesosphere
Apache Mesos and DC/OS allows users to deploy distributed applications and in particular micro-services across a large cluster. Therefore, networking becomes an important aspect especially when trying to provide highly-available applications on top of an unreliable infrastructure.
In this talk, we will first present the various challenges around networking for distributed micro-service architectures, including
* Connectivity
* Service Discovery
* Load-balancing
* Isolation

As for most of the above challenges there is not a one-size-fits-all solution we have an in-depth look at the trade-offs between different solutions.
Afterwards, we will deep dive into the actual implementation of the different components in order to understand how we can achieve a scalable networki

Speakers
avatar for Jörg Schad

Jörg Schad

Head of Machine Learning, ArangoDB
Jörg Schad is Head of Machine Learning at ArangoDB. In a previous life, he worked on built machine learning pipelines in healthcare, distributed systems at Mesosphere, and in-memory databases, conducted research in the Hadoop and Cloud area. He’s a frequent speaker at meetups... Read More →
avatar for Avinash Sridharan

Avinash Sridharan

Software Engineer, Mesosphere
Love containers, especially networking them !! Have worked (and studied) in the field of networking for the past decade or so. Very interested in designing and developing new network technologies. Want everything to be fast, distributed, and extremely efficient.


Wednesday May 31, 2017 16:50 - 17:30
Room 1

16:50

Fast Releasing and Testing of Gentoo Kernel Packages and Future Plans of the Gentoo Kernel Project - Alice Ferrazzi, Gentoo Kernel Project
When I joined the Gentoo Kernel Team, my initial focus has been on taking a closer look at the kernel releasing process and documenting it.  My investigations revealed that there was a discrepancy between the release steps followed by different developers.

As a result, I proposed a policy to follow a more standardized approach to making releases.
My study concluded that our team can automate kernel package releasing and testing, which frees up developer time to work on writing patches and fixing bugs.

The Gentoo Kernel CI has substantially decreased the time and human effort required to make kernel releases and enabled us to keep up with the upstream release schedule - something that wasn't feasible before the continuous integration. I will talk also about Gentoo Kernel Project future plans.

Speakers
avatar for Alice Ferrazzi

Alice Ferrazzi

OSS開発者, サイバートラスト株式会社
Alice Ferrazzi is a Gentoo Linux Developer and the Gentoo Kernel Project Leader, working on Gentoo ebuild, eclass writing and kernel. She is also part of the Gentoo Foundation Board Members. She holds Gentoo study meetings in Tokyo, Japan and organizes Gentoo booth at various open... Read More →



Wednesday May 31, 2017 16:50 - 17:30
Hall B-1

16:50

The Many Approaches to Real-time and Safety Critical Linux Systems - Wolfgang Mauerer, Technical University Regensburg/Siemens AG
Linux and Real-Time have become a widespread combination that is deployed in many industrial solutions. Real-Time requirements are often combined with safety requirements, and satisfying both is only possible when the whole system architecture is designed with both goals in mind, which goes well beyond just applying the preempt_rt or Xenomai patch sets. Particular attention in this talk is given to partitioning systems into critical and uncritical components, which has gained substantial attraction with the advent of multi-core CPUs in the embedded segment. In the talk, we discuss possible architectural approaches to safety-critical real-time Linux systems, and highlight their advantages and disadvantages. We also provide guidelines on which architectural option is suited best for which appliances and use-cases.

Speakers
WM

Wolfgang Mauerer

Senior Key Expert/Professor, Siemens AG
Wolfgang Mauerer is a professor of theoretical computer science at the Technical University Regensburg, and a senior key expert at Siemens Corporate Research, Competence Centre Embedded Linux. He serves on the technical steering committee of the Linux Foundation's Civil Infrastructure... Read More →


Wednesday May 31, 2017 16:50 - 17:30
Private Dining

16:50

The Rise of the Copyright Troll - Shane Coughlan, OpenChain Project
This talk will explore the rise of alleged copyright “trolling” around Open Source. It will explore what has happened, how organizations have reacted, and what is likely to occur next. The focus will be on lessons learned and how these lessons can be applied to real-world commercial situations.

Speakers
avatar for Shane Coughlan

Shane Coughlan

OpenChain General Manager, Linux Foundation
Shane Coughlan is an expert in communication, security and business development. His professional accomplishments include spearheading the licensing team that elevated Open Invention Network into the largest patent non-aggression community in history, establishing the leading professional... Read More →



Wednesday May 31, 2017 16:50 - 17:30
Room 6
 
Thursday, June 1
 

11:00

WalB: Real-time and Incremental Backup System for Block Devices - Kota Uchida, Cybozu, Inc.

WalB is an open-source backup system that consists of block devices, called WalB devices, and userland utilities, called WalB tools. A WalB device records write-I/Os. WalB tools extracts them to create restorable snapshots in an incremental manner.

Compared with dm-snap and dm-thin, WalB is designed to achieve small I/O latency overhead and short backup time. We conducted an experiment to take an incremental backup of a volume under random write workload. The result confirms those advantages of WalB.

Cybozu cloud platform, which has 500TB volumes and processes 25TB write-I/Os per day, is required to achieve (1) stable workload performance without I/O spikes which may affect application user experience and (2) short backup interval specified in our service level objective. WalB satisfies the requirements, while dm-snap is not enough to and dm-thin is not expected to.


Speakers
avatar for Kota Uchida

Kota Uchida

Software Engineer, Cybozu, Inc.
Kota UCHIDA works at Cybozu, Inc. He has earned his Bachelor's and Master's degrees in Engineering at Tokyo Institute of Technology. He has talked about his research at Foundation on Software Engineering 2012 (FOSE 2012). He has published a book about Intel x86 architecture and an... Read More →



Thursday June 1, 2017 11:00 - 11:40
Private Dining

11:50

OSS CVE Trends - Kazuki Omo, Secure OSS SIG & SIOS Technology Inc.
Nowadays security incident is increasing more and more. Then lots of vendor/community/institute are making efforts to find vulnerability on software. Not only commercial software, but also OSS is having vulnerability (remember HeartBleed, DirtyCow, and so on). Then lots of security researchers are reporting vulnerability and publish it with CVE-ids which is assigned by MITRE.

In this Presentation, Kazuki Omo will report recently trends of OSS CVE (2015-2017), and tell what kind of vulnerability are focused nowadays and how you can get vulnerability information quickly.

Speakers
avatar for Kazuki Omo

Kazuki Omo

Executive Officer, SIOS Technology Inc.
Over 20 years experience in Unix/Linux/Windows system and many of Security related product. Working for OSS community over 15 years. - Published SELinux and related security articles from 2004-2018. - Presentation on Open Source Summit Japan 2017 "OSS CVE Trends". - Presentation on... Read More →



Thursday June 1, 2017 11:50 - 12:30
Room 6

14:00

AER Functionality of Pass-Through PCI-e Device in Qemu - Jin Cao, Fujitsu
In Qemu, assigned PCI devices are go through vfio_pci now, but one of important features still is not implemented: AER error recovery, due to its complexity. When error happens, it causes a Qemu VM to stop. A single pass-through device's error leads to a whole VM failure, that is not quite reasonable.

Fujitsu team focused on this topic two years ago, but Qemu community doesn't have a mature solution for this problem until now. Fujitsu has sent a dozens of patchsets, finally got some agreement and made some
achievement.

In this presentation, Cao jin will review the two phases of development, introduce the functionality arch of the feature, the difficult problem found in development, and current status.

Speakers
avatar for Cao Jin

Cao Jin

Engineer, Fujitsu
Cao jin is linux kernel developer now employed by Fujitsu. He worked for Huawei & Alcatel-lucent before. He contributed almost 100 patches during the first year in open source world(qemu, kernel, etc). He has strong interests in the low level things.Half a year ago, he began to dive... Read More →


Thursday June 1, 2017 14:00 - 14:40
Private Dining

14:00

The Kubernetes API & Next Generation Automation Tools - Ian Lewis, Google
The Kubernetes API provides new ways to build automation tools that are highly available and scale to serve clusters with many thousands of nodes. During this session attendees will learn how to move beyond shell scripts and configuration management tools and leverage cluster level APIs and distributed systems design patterns to build next generation automation tools. Attendees will learn how to extend the Kubernetes API and utilize their own custom declarative configuration to build controllers that automate job scheduling or schedulers that allow you to place jobs based on your rules.

Speakers
avatar for Ian Lewis

Ian Lewis

Developer Advocate, Google
Ian is a software engineer at Google and contributor to the gVisor project. Ian has had various developer and operations roles throughout his career and enjoys working in environments with diverse ways of thinking. Ian has been living in Tokyo since 2006 and is active in the open-source... Read More →


Thursday June 1, 2017 14:00 - 14:40
Room 6

14:50

Parallelizing CI Using Docker Swarm-Mode - Akihiro Suda, NTT Corporation
Slowness of CI is a critical issue in software development, because it discourages engineers from writing tests, and hence deteriorates the quality of the product.

In this presentation, Akihiro Suda will talk about how to accelerate CI by executing test functions in parallel, across a Docker Swarm-mode cluster.
One of the major challenges was the nonuniformity of the makespan. e.g. some chunk of test functions can take 30 minutes, some chunk can take just 10 seconds...
So, he mitigates such a nonuniformity by randomizing the composition of chunks of test functions.

As a result, for example, the integration test of Docker itself that had taken more than 80 minutes can be finished in 4 minutes, with 10 Docker Swarm-mode nodes.
This hack can be easily applied to CI of other software as well.

Speakers
avatar for Akihiro Suda

Akihiro Suda

Software Engineer, NTT
Akihiro Suda is a software engineer at NTT Corporation, a Japan-based telecommunication company. He has been a core maintainer of Moby (former Docker Engine) since November 2016. He has been also a maintainer of several opensource container software such as CNCF containerd and Moby... Read More →


Thursday June 1, 2017 14:50 - 15:30
Room 6

16:00

Adding New CPU Architecture to QEMU - Marek Vašut, DENX Software Engineering
This talk provides an overview of the internals of the QEMU system emulator. After a brief introduction of QEMU itself, Marek will explain what pieces need to be filled in to add a new architecture to QEMU. The first topic is TCG, and how it is used to emulate the foreign instruction set on a host system with minimum performance penalty. Since emulating instructions properly has various pitfalls, Marek will point those out in the implementation of the Intel NiosII CPU instruction set for QEMU, which is tiny yet already capable of running Linux. Still, emulating instruction set does not make a usable emulation, thus the second part of the talk will focus on implementing models of real hardware in the context of QEMU. Finally, Marek will explain how to add support for emulating Linux userspace binaries.

Speakers
avatar for Marek Vasut

Marek Vasut

Software engineer, Self employed
I have been a contractor for multiple companies for many years. My primary responsibility is designing and implementing customer-specific functionality. One important aspect of my work is leveraging the benefits of working inside the mainline Linux, U-Boot and OE / Yocto Project... Read More →


Thursday June 1, 2017 16:00 - 16:40
Private Dining

16:50

BoF: SELinux Policy/Module Developer's - Yuichi Nakamura, Secure OSS SIG & Hitachi
Analyzing, modifying and appending security policy of SELinux is not easy, because SELinux provides very fine grained access control and there are so many rules. In order to facilitate these tasks, there are various tools handling SELinux policy, but they are not enough.

In this BoF session, existing tools are reviewed, and problems of them are shared at first. Secondly, requirements to improve tools are discussed with participants.

Speakers
YN

Yuichi Nakamura

Senior Engineer, Hitachi,Ltd
Yuichi Nakamura works for Open Source Solution Center at Hitachi,Ltd, and is responsible for security solution. He received a PhD degree from Okayama University about a study including SELinux policy development. He gave presentations in many OSS events such as OLS and ELC. He also... Read More →


Thursday June 1, 2017 16:50 - 17:30
Restaurant Carnelian

16:50

Automating Workflows for Analytics Pipelines - Sadayuki Furuhashi, Treasure Data
Learn how to leverage new workflow management tools to simplify complex data pipelines and ETL jobs spanning heterogeneous systems. In this technical deep dive from Treasure Data, company founder and chief architect walks through the codebase of Digdag, our recently open-sourced workflow management project. I’ll show how workflows can break large, error-prone SQL statements into smaller blocks that are easier to maintain and reuse. I also demonstrate how a system using ‘last good’ checkpoints can save hours of computation when restarting failed jobs and how to use the workflows to automate data lifecycle management across Apache Hadoop, PostgreSQL, Amazon S3 and Apache Spark. You'll see a few examples where SQL-as-pipeline-code gives data scientists both the right level of ownership over production processes and a comfortable abstraction from the underlying execution engines.

Speakers
avatar for Sadayuki Furuhashi

Sadayuki Furuhashi

Founder and Software Architect, Treasure Data
Sada is the original author of Fluentd, Embulk, MessagePack, and now Digdag: an open-sourced workflow management project. Sada is a co-founder Treasure Data, Inc., a cloud-based data warehousing and analytics service. He has been working on production distributed systems for a decade... Read More →


Thursday June 1, 2017 16:50 - 17:30
Private Dining

16:50

Bare Metal Container - Kuniyasu Suzaki, AIST
Container technology becomes popular because of easy customization and quick execution. However, container does not allow kernel customization. For example, DPDK cannot run on Docker, because DPDK requires “igb_uio” and “rte_kni” kernel modules.
BMC: Bare Metal Container solves this problem by a mechanism that offers a suitable kernel for a container image and boots them on a remote physical machine. Although BMC requires the overhead of booting, applications improve their performance and compensate the overhead.
BMC is compatible to Docker and reuses the container images offered by DockerHub. BMC also offers a repository for kernel images. Users select a suitable one for their application. The experiments showed the effect of kernel optimizations for CPU (Hyper Threading), memory (Transparent Huge Pages) and network (Receive Flow Steering) from a low power Atom to a high speed Xeon.

Speakers
KS

Kuniyasu Suzaki

Senior Researcher, AIST
He is a senior researcher at Information Technology Research Institute (ITRI) of National Institute of Advanced Industrial Science and Technology (AIST). His research interests are high performance computing and security. He made presentations at CodeBlue'14, BlackHat SaoPaulo'14... Read More →


Thursday June 1, 2017 16:50 - 17:30
Room 6
 
Friday, June 2
 

11:00

Advanced Continuous Delivery Strategies for Containerised Applications Using DC/OS - Jörg Schad, Mesosphere
Using a container orchestration platform like the Datacenter Operating System (DC/OS) makes it trivial to setup an automated continuous deployment pipeline that pushes code to production on every commit (perhaps with some tests thrown in the middle). This is a win for customers (they see new features sooner), developers (much less bureaucracy with each release) and operators (fewer changes with each release means less risk).

In this presentation, we'll introduce DC/OS, an open source distributed operating system and container orchestrator based on the production proven Apache Mesos. We will then describe and demonstrate advanced deployment strategies including canary deployments and blue/green deployments, showing you how you can integrate these with continuous deployment pipelines on DC/OS to perform advanced automated deployments with low risk over thousands of machines.

Speakers
avatar for Jörg Schad

Jörg Schad

Head of Machine Learning, ArangoDB
Jörg Schad is Head of Machine Learning at ArangoDB. In a previous life, he worked on built machine learning pipelines in healthcare, distributed systems at Mesosphere, and in-memory databases, conducted research in the Hadoop and Cloud area. He’s a frequent speaker at meetups... Read More →


Friday June 2, 2017 11:00 - 11:40
Room 6

11:00

Container Interfaces for Storage - Are We There Yet? - James Bottomley, IBM Research
Many talks about containers start with Orchestration systems like Docker or Kubernetes. However, this one will look at the storage impacts on the actual in-kernel container API. With the addition of the superblock namespace (essentially a user namespace for the kernel to filesystem boundary) much of the stage is now set for fixing one of the biggest underlying container problems: that of translating unprivileged container writes into real filesystem uid/gids. This talk will examine how this system works, why it is necessary and what pieces still need to be added for orchestration systems to make use of it (yes, we'll also cover fully unprivileged Docker ... but only briefly).

Speakers
avatar for James Bottomley

James Bottomley

Distinguished Engineer, IBM
James Bottomley is a Distinguished Engineer at IBM Research where he works on Cloud and Container technology. He is also Linux Kernel maintainer of the SCSI subsystem. He has been a Director on the Board of the Linux Foundation and Chair of its Technical Advisory Board. He went to... Read More →


Friday June 2, 2017 11:00 - 11:40
Room 1

14:50

MINCS - Mini-Container Shell Scripts - Masami Hiramatsu, Linaro
MINCS is a shell script based container engine, which will be good for learning how the container is made on Linux. It is also good for resource-limited environment because it is a set of 20KB scripts and can run on busybox. This talk will show how to use MINCS with Docker image and debootstrap rootfs, and how it is implemented by using shell commands. It also includes recent MINCS updates which adding cross-arch support by qemu-system and qemu-user so that you can easily build an emulation environment for testing cross-arch applications and kernel.

Speakers
MH

Masami Hiramatsu

Tech Lead, Linaro
Masami Hiramatsu is a linux kernel maintainer and working for Linaro Ltd., as a tech lead. He has been working on kprobes and dynamic event tracing parts like perf-probe, ftrace, etc.


Friday June 2, 2017 14:50 - 15:30
Room 1

14:50

Unifying Android and Mainline Kernel Graphics Stack - Gustavo Padovan, Collabora Ltd.
The Android ecosystem has tons of out-of-tree patches and a good part of them are to support Graphics drivers. This happened because the Upstream Kernel didn't support everything that is needed by Android. However the Mainline Graphics Stack has evolved in the last few years and features like Atomic Modesetting and Explicit Fencing support are making the dream of running Android on top of it possible. In other words, we will have Android and Mainline Kernels sharing the same Graphics stack!

This talk will cover what has been happening both on Android and Mainline Graphics Stacks in order to get Android to use the Upstream Kernel by default, going from what Android have developed to workaround the lack of upstream support to the latest improvements on the Mainline Graphics Stack and how they will fit together.

Speakers
avatar for Gustavo Padovan

Gustavo Padovan

Software Engineer, Collabora
Gustavo Padovan holds a BSc. Computer Science from the University of Campinas, Brazil. He is Linux Kernel Developer and works at the open-source consultancy Collabora Ltd. In the Kernel he has worked in a number of areas, notably as Maintainer of the Bluetooth Subsystem and has been... Read More →


Friday June 2, 2017 14:50 - 15:30
Hall B-1

16:00

Taming the Container Security Beast - Tim Mackey, Black Duck Software
Container orchestration solutions introduce a level of security complexity into the lifecycle of an application. Continuous deployment of container images is fundamentally challenged by the rate of security disclosures. Understanding if a vulnerable image exists, what the vulnerabilities within an image might be and where the images are deployed is a daunting task. One traditional response is to invest in perimeter defenses, but what happens when you don’t own or control the perimeter? Taking a step back, we realize the applications and dependencies are what’s under attack. Having a clearly defined security model covering development, staging and deployment is required. That security model also needs to take into account the vulnerability lifecycle from defect discovery through patch creation with a focus on when malicious actors have an advantage.

Speakers
avatar for Tim Mackey

Tim Mackey

Senior Technical Evangelist, Black Duck by Synopsys
Tim Mackey is a technology evangelist for Black Duck Software specializing in the secure deployment of applications using virtualization, cloud and container technologies. Prior to joining Black Duck, Tim was most recently the community manager for XenServer and was part of the Citrix... Read More →


Friday June 2, 2017 16:00 - 16:40
Room 6

16:50

High performance Microservices on Linux with Apache Thrift - Randy Abernethy & Aki Sukagawa, RX-M, LLC
The world is rapidly adopting cloud native approaches to software development, reaping the synergies produced by combining microservices, Linux container technology and dynamic application orchestration. Containers offer performance gains and simplified operations, however maximizing microservice performance involves API level considerations. In this demonstration and talk we will take a look at how Apache Thrift can be used to generate material performance gains in container packaged microservices. We'll contrast Apache Thrift with REST and gRPC, describing the strengths and weaknesses of each. We'll also demonstrate how to seamlessly package Apache Thrift microservices with Docker and how to scale them with Kubernetes on a Linux cluster. The talk will conclude with a demonstration of microservice interface evolution and rolling upgrades.

Speakers
RA

Randy Abernethy

Managing Partner, RX-M, LLC
Tech Entrepreneur, coder, startup adviser, financial technology pioneer, Apache Thrift committer, author and highly experienced Destiny guardian.



Friday June 2, 2017 16:50 - 17:30
Hall B-1

16:50

Intelligent Storage: SQL Execution on GPU Closely Connected with SSD - KaiGai Kohei, PG-Strom Development Team
KaiGai Kohei presents a new feature built on PostgreSQL to accelerate analytic queries, with leverage of heterogeneous hardware. PostgreSQL has already allowed off-loading some CPU intensive SQL workloads onto GPU device, to utilize its multi-thousands cores to process tons of records. Not only massive parallel computing, GPU also has an interesting feature which runs peer-to-peer DMA from other PCIe device to GPU. It makes possible to load the data blocks of PostgreSQL on NVMe-SSD to GPU directly. Once data blocks get loaded, we already have a mechanism to run WHERE/JOIN/GROUP BY on GPU, then write back its execution results to host RAM. It is usually much smaller than raw data, thus less amount of records CPU must process. From the standpoint of applications, a couple of SSD and GPU look like an intelligent storage that understand and run SQL then returns minimum necessary data set.

Speakers
avatar for KaiGai Kohei

KaiGai Kohei

Chief Architect, PG-Strom Development Team
KaiGai Kohei is lead developer of the PG-Strom project. He has more than ten years experiences around open source software development of Linux, PostgreSQL and so on. Especially, he has contributed some of core functionality of PostgreSQL on the area of security enhancement and pluggable... Read More →


Friday June 2, 2017 16:50 - 17:30
Room 6