The hottest wave in Copenhagen kubecon will push c

  • Detail

Where does the huge wave of kubecon in Copenhagen push cloud computing

in May, kubecon + cloudnativecon Europe 2018, one of the most influential technology summits in the container field, was held in Copenhagen, Denmark. Huawei cloud, which holds the dual seats of kubernetes Steering Committee and CNCF's Technical Supervision Committee, was invited to share their technical experience in kubernetes, security, microservices, machine learning, serverless, Devops and other fields with other IT giants, and discuss and prospect the present and future of cloud native technology together

in addition to technology sharing, this conference also attaches great importance to customer cases, including spotify, Wikipedia, youtube, Adidas, financial times, eBay, Norwegian tax administration center and other enterprise customers. The accuracy differentiation of hydraulic universal testing machine will come to share how they use containers and their surrounding technologies in the production environment

kubernetes commercial maturity

cloudy has become an inevitable trend

kubernetes, as the core project of CNCF, is also the first project to successfully enter the commercial ready, and the sharing of its production practice has become a major focus of this conference


on the first day of keynote, the production practice cases from the European Center for Nuclear Research (CERN) surprised the audience

as the largest particle physics research center in the world, CERN has a huge computing demand. In a self built data center, CERN has built more than 210 k8s clusters to schedule and manage the infrastructure with 320000 cores and more than 10000 hypervisors. The deployment scale of these clusters varies. The smallest one is only dozens of nodes, while the largest one has reached thousands of nodes

to facilitate the unified management of workloads in these clusters, CERN uses k8s Federation as a unified platform entry. At the same time, CERN has also created k8s clusters on cloud platforms such as Kom cloud, Google cloud, azure and AWS, which are Huawei's partner public cloud open tele, and connected them to their platforms, so as to quickly respond to the soaring computing volume during large-scale events such as technology summits

experts from CERN share the case of using cluster Federation to manage k8s multiple clusters

creating k8s clusters on two or more cloud platforms and deploying workloads has become a routine practice for many k8s adopters. Compared with the past, users can deploy services on cloud platforms at the same time relatively easily, and enjoy the advantages of different cloud platforms

it is not difficult to find that at present, the container service based on kubernetes has almost become the standard service of various cloud platforms. Thanks to the promotion of k8s software consistency certification project, more and more manufacturers will provide certified k8s distribution as a basic requirement. Cloudy support has become an inevitable trend. With the development of cloudnative, we believe that in the near future, the cloud native platform with k8s as the core will truly realize cloud agnostic. Users can really easily realize the free migration of workload across clouds and clusters

core and basic problems have been solved

how to eliminate security anxiety in the context of cloud native

the past two years are the entrepreneurial period of CNCF. The community takes kubernetes and container technology as the core of the platform, complements capabilities in the fields of observability, operability, microservice discovery, and builds a flexible and scalable basic platform

with the increasing adoption, implementation and operation of container, microservice and other technologies in the production environment, more and more people pay attention to the security issues in the context of emerging technologies. How to eliminate these concerns is also the focus of CNCF's next efforts


runtime, Google has brought their own secure container solution gvisor. Gvisor provides a new Sandbox container runtime environment, which can provide a similar isolation effect with virtual machines while ensuring the advantages of lightweight

gvisor provides isolation boundaries by intercepting application system calls in user space and acting as a guest kernel. In addition, gvisor does not need fixed resources, and it can adapt to changing resource conditions at any time

gvisor project provides new ideas for container safety and enriches the ecosystem of safe container technology. Although there is still a way to go before commercial use, it will undoubtedly bring great help in the future in terms of promoting safe containers to the mainstream market

kubeflow released version 0.1

significantly reduced the deployment threshold of machine learning framework

in recent years, the development of machine learning can be described as rapid progress. How to give full play to the advantages of kubernetes and use it as a deployment platform to provide a convenient and scalable machine learning framework is one of the key topics. The initiation of kubeflow project is an attempt to find the simplest open source solution


since kubecon + cloudnativecon in North America announced the establishment of the project last year, kubeflow has attracted more than 70 contributors from more than 20 organizations, including Google, Microsoft, RedHat, Huawei, Alibaba cloud, etc., contributed more than 700 code submissions, and obtained a star of 3.1k, with the fastest growth rate leaping to the top 2% of GitHub

version 0.1 released this time provides a set of the most concise software package, which is convenient for users to develop, train and deploy machine learning framework

in the next few months, the kubeflow project will be committed to the release of 0.2. By the end of this year, after the release of version 1.0 of kubeflow, the kubeflow project will seek a formal governance community, hosted under CNCF or other communities

service mesh continues to be popular

istio leads the tide

enterprise cloud, container and microservice are two powerful tools. The container shields the application's perception of the environment, simplifies the consistency of software package distribution, and avoids repeated work. The architecture of turning to micro services shields the application's perception of services, and enables the business team to focus more on its own professional fields. As for how to do load balancing, fusing, telemetering and other issues, they can be removed through service mesh. A high degree of focus can greatly improve productivity


istio is the most promising project to become the de facto standard of service mesh. Since its release in May last year, it has become popular rapidly with the deep integration, zero invasion, easy expansion and other advantages with k8s, supported by major manufacturers such as Google and IBM. In less than a year, it won 8000 + stars, attracted nearly 200 contributors to participate in code development, and became the hottest project in the k8s ecosystem since last year

a major topic of istio in this conference is cloudy and multi cluster. K8s provides a consistent platform environment between multi clouds, but how to realize cross cloud and cross cluster service discovery and flow control has been pending

in the k8s Federation project, there is a simple version to realize the priority routing of this cluster. Istio provides the feature of multi cluster support in the latest version 0.8, which complements the service management capability of k8s in the multi cluster scenario

serverless field release events

standard cloudevents 0.1

with the development and widespread adoption of cloud technology, applications become more and more decentralized, the number of integrations continues to grow, and the concept of event driven architecture and serverless rises

in order to solve the interoperability problem of serverless, the CNCF serverless working group has been committed to the development of cloudevents, the serverless event Standard Specification, since the white paper was completed at the end of last year. Open source players Huawei, Google, Microsoft, IBM, red hat, etc. have actively invested in it and made great contributions to the project

the scope of cloudevents 0.1 released at this conference is very simple:

provides a consistent set of metadata that can be included in event data, making events easier to apply to publishers, middleware, subscribers and applications. In short, it is a standard event envelope

cloudevents' General Metadata makes events easier to route, fan out, track, replay, and basically maintain. They are more portable, smoother and easier to transmit across platforms. At present, network bandwidth, cost and latency are still the main challenges, but cloudevents' simple metadata definition can bring good portability to data in many scenarios

focus on k8s to accelerate innovation

cloudnative programming framework came into being

as we all know, kubernetes achieved loose coupling in architecture at the beginning of design, and then made many plug-in frameworks and scalability improvements and enhancements in the evolution. The introduction of the concept of operator has standardized a large number of scenarios that have customized expansion requirements for k8s

however, there is still a certain threshold for the development, testing, operation and maintenance of the operator itself. The operator development framework aims to summarize the excellent practical experience in the existing implementation and form a set of standards to help reduce the threshold of application development, testing and operation and maintenance on k8s. This is a great blessing for enterprise users who suffer from the difficulty of operation and maintenance after business transformation and have customization needs for k8s


this time kubecon also brought a very novel project ballerina. This is a cloud native programming language for integration

Ballerina developers believe that applications written by people in the future will increasingly rely on APIs, and integration is an important specification to open up elastic communication between endpoints

therefore, they integrated the basic concepts of distributed system integration into the language, designed this compiled, transactional, static strongly typed programming language, and provided a type safe concurrent environment

Ballerina supports text and graphic syntax. In addition to conventional text coding, developers can also organize code by charts in the designer. This further reduces the development threshold of cloud native applications, and developers can easily implement micro services with distributed transactions, reliable messaging, flow processing and workflow

the curtain opens

the real good play of installing and checking electrical systems has just begun

for distributed application development, we see Erlang, fault-tolerant programming and development framework, etc., which are more bound to different programming communities and software stacks

we are glad to see that kubernetes, with its many amazing features and huge ecological power, has entered a development process similar to the commercialization of many distributed systems, becoming increasingly suitable for daily developers and enterprises, and finally becoming a horizontal technology widely used in the industry

the use threshold of k8s will change from white box to black box. Developers do not need to master too much knowledge of k8s. They can realize distributed system style programming based on a set of standardized primitives

people don't have to worry about how to compile the code, how to build the image, and how the test is different from the configuration of the production environment. A common code submission action can trigger the delivery pipeline of the whole process from compilation, construction, testing to production, operation and maintenance. The rollback in case of problems will also be so simple, because the operation of code submission is atomic

looking back at the moment, maybe many people will think that k8s has gradually stabilized and become boring, but looking at the whole cloudnative ecosystem, many new and interesting projects are springing up, but among them, what deserves the attention of enterprises is: 1. The growth rate of automobile production and sales, which has the largest weight in the total industrial output, is picking up; 2. Key points of machinery industry

Copyright © 2011 JIN SHI