What did Kubecon say the first day?

[Editor's Note] David O'Dwyer wrote for Kubecon 2015 , highlighting some of the latest features of the Kubernetes v1.1 version, and summarizing the first day's speeches, allowing us to develop Kubernetes, Ecology has a more step to understand.

@Container Container Technology Conference will be held on January 24, 2016 in Beijing, from the Philharmonic , microblogging, Tencent, where to network, the United States Mission Cloud, Jingdong, mushroom Street, Hewlett-Packard, Rush cartoon and other well-known companies responsible for the technology People will share their container application case.

This week the Kubecon 2015 conference is going on, so there is really no more time to try Kubernetes anymore. If you have never heard of it, you can always read our intro or official website . If you are interested, you can start a local cluster through Vagrant and try it out!

Big news

Kubernetes v1.1 has been released! This version brings some exciting new features, some of the API has been extended in the API, until now the introduction of a stable version:

  • Pod Horizontal Auto Extension – When a specified resource limit is reached, such as CPU usage, you can dynamically add a specified copy of the Pod.
  • Job object – Respond to batch and other one-time workloads.
  • HTTP Load Balancing – Now you can use an ingress controller to distribute HTTP requests directly to different services on different domain / path load balancing.
  • Kubectl tool improvements , including the ability to run containers interactively on a cluster for debugging and improved scrolling updates.
  • A series of bug fixes , see the GitHub page .

first day

  • Brendan Burns is a senior engineer at Kubernetes, who demonstrated a zero-stop rolling update demo – providing 1 million requests per second while updating. Really shocking!
  • Several speeches have referred to the application of Kubernetes in the production environment, starting with the build system of build.golang.org to Deis's lecture on running the Kubernetes-based PaaS system, and eBay's article on running Kubernetes in OpenStack.
  • Other presentations focus on the pluggable version of the Kubernetes component. Eugene Yakubovich, a Core OS flannel defender, spoke about the container network interface . The focus of this lecture is to try to standardize the container network solution (where Kubernetes relies on it to implement pod internal communication) and how do you add a Kubernetes web plugin.
  • As Sysdig has done a similar lecture at the London container conference, we have had some understanding. This time the Kubernetes integration, which allows you to filter based on Kubernetes objects such as namespaces and services. I just want to say that they already have a very cool container monitor dashboard, but our interface based on ncurses will make people more impressive, at least the audience's response is very positive.
  • There is also a speech that draws interest to Pachyderm , a containerized data store for large datasets. It extends the container-group commit-based, btrfs copy-on-write file system. Is still in the early stages, but at some point in the future may be another alternative to Hadoop!

Today is the end of a " Ubernetes " speech, the theme is about how to cross the multiple data centers and suppliers to integrate Kubernetes cluster, how to solve data and application migration, global expansion and flexibility to deal with major failures? The lecture addresses the details of the implementation details and Kubernetes in the next few months to solve these problems in the road map. Of particular concern is the deployment of Ubernetes to avoid potential suppliers to lock the underlying infrastructure, because now your supplier range is already very large. In addition, how to adjust the configuration of your application components to achieve optimal utilization for each vendor platform.
Original link: KUBECON: DAY 1 (translation: Zhu University)

Heads up! This alert needs your attention, but it's not super important.