Author: your heart, into super, Yuan Yi, the heart of water, Zhong source, soldiers wash
Kubernetes v1.16 released
Concern some of the features in this release and Feature:
GA CRD officially entered the stage;
Admission Webhook GA officially entered the stage;
A series of enhancements and fixes, and Volume of CSI.
Ants gold dress open source project ElasticDL
September 11, 2019 ants gold dress on the Google Developers Conference in Shanghai station ElasticDL open source project, which is the industry’s first system based on open source to achieve flexibility TensorFlow depth learning. ElasticDL achieve fault tolerance and flexibility of distributed scheduling deep learning, can greatly improve the overall utilization of the cluster, while significantly reducing the time a user submits a job after waiting for a job to start (pending time).
Oracle announced the permanent free part of a cloud computing service
Oracle CEO Larry Ellison announced at OpenWorld conference to launch the Oracle Cloud Free Tier, large and small companies and organizations, developers, students and educators can build, learn and explore the full functionality of Oracle Database and Oracle autonomy cloud infrastructure.
The introduction of Oracle Cloud Free Tier (free package plan) There are two components, namely the free trial program and permanent free cloud services.
Significant progress upstream
Repair API Server access Webhook flow imbalance.
In cluster size or QPS higher cluster, it is recommended to avoid Webhook Port this repair work caused by a single point of performance bottleneck: https: //github.com/kubernetes/kubernetes/pull/82090
Repair Bug Pod toleration caused.
Repair Pod toleration there are a plurality, causing comparison & merging the Bug. Port recommend this fix in order to improve the robustness of the scheduling Pod semantics: https://github.com/kubernetes/kubernetes/pull/81732/files
KEP: reduce API Server to Watch event sequencing overhead.
API Server is currently under a large-scale cluster, there are a lot of CPU time slice and resources are used for serialization and de-serialization, in order to further improve cluster performance and throughput, we should think of ways to optimize serialization and de-serialization problem. https://github.com/kubernetes/enhancements/blob/master/keps/sig-api-machinery/20190329-less-object-serializations.md
Discuss the principles issued Knative
Set the minimum K8s supported version;
Only supports four Knative recent version;
After the announcement of date, knative beta api version to support a minimum of nine months, GA version supports at least 12 months. For v1alpha1 as a beta version;
Runtime API and need a recent version of the four tested for compatibility.
- Dead-Letter Channel Design
Dead letter queue for the design, is expected to be released in Eventing v0.10.0 release.
- Update on Performance Task Force
Eventing performance test conducted baseline test throughput, i.e. without the use of index data in the broker or channel. The next step will be based on the throughput test broker or channel.
Open source projects recommended
Google on codeview guidance document. Google has long been heard for a good codeview culture, as engineers, we should learn codeview culture to achieve better collaboration and co-workers. (Chinese version Reference: https: //jimmysong.io/eng-practices/)
User Experience great Kuberntetes of gitops CI / CD-based system – argo-cd.
Ants gold dress open source distributed depth learning system based on the elastic TensorFlow of ElasticDL.
This week Recommended Reading
“Attack on Java – native cloud transformation.”
The advent of native cloud era, and Java developers in the end What is the connection? It is said that cloud they not native to Java exist. However, the authors of this paper thought that the era of cloud native, Java still be competent role of “giant”. The authors hope through a series of experiments, the students open up horizons, provide useful thinking.
“PaaS and KaaS What is the difference? Important When? “
PaaS and KaaS offers many similar features, so it is easy to confuse the system as a service and platform as a service Kubernetes, but they are essentially different, if you choose the wrong option, which could have serious impact. In this webinar, Mirantis of Nick Chase explains the difference between the two, and how to choose a better way for your particular use case.
“Kubernetes authentication and authorization operations Raiders: Access Control of Service Account”
Depth study access service account control.
“Behind Kubernetes continued rise: security issues are still, life cycle doubt”
To better understand the safety and use of container and Kubernetes trend, the article incorporates nearly 400 members of the findings from the security, DevOps and product team, would like to know how the various organizations using container technology, security issues currently facing and the further development specific insights.
“Using Django, Prometheus, and custom applications Kubernetes targets”
The article emphasizes the importance of the application of custom indicators, with code examples demonstrate how to design and integrate indicators Prometheus to Django project, provides a reference for developers to build applications using Django’s.
“Building Kubernetes cluster – Select node number and size of work”
The article describes the dimension of the respective advantages and disadvantages of using less large and more small node to node cluster formation Kubernetes, and practical experience provides a general method of selecting node number and size of work.
Alibaba Cloud native micro-channel public number (ID: Alicloudnative) focus on micro service, Serverless, container, Service Mesh and other technical fields, focusing popular technology trends in cloud native, cloud native large-scale landing practice, do most understand cloud native developers the public.