SIG-Node, with Dawn Chen

Kubernetes Podcast from Google

Episode | Podcast

Date: Wed, 26 Sep 2018 05:35:00 +0000

<p>Dawn Chen, TL for SIG-Node and the Google Kubernetes Engine node team, joins <a href="">Craig and Adam</a> this week. She has worked on containers and container schedulers since 2007 - not a typo. We also bring you the news, in part from the echo chamber of Google Cloud Summit in Sydney.</p> <p>Do you have something cool to share? Some questions? Let us know:</p> <ul> <li>web: <a href=""></a></li> <li>mail: <a href=""></a></li> <li>twitter: <a href="">@kubernetespod</a></li> </ul> <h3 id="chatter-of-the-week">Chatter of the week</h3> <ul> <li><a href="">Stickers!</a></li> <li><a href="">Google Cloud Summit in Hong Kong</a></li> <li><a href="">Google Cloud Next in London</a></li> <li><a href="">Gartner Symposium Orlando</a></li> <li><a href=""> KubeCon Shanghai</a></li> </ul> <h3 id="news-of-the-week">News of the week</h3> <ul> <li><a href=""> NetApp acquires StackPointCloud</a></li> <li>Cloud news: <ul> <li><a href="">Sandbox pods on GKE</a> <ul> <li><a href=""> Signup form</a></li> </ul> </li> <li><a href=""> Kubernetes tools for Azure Stack</a></li> <li><a href=""> EKS can generate kubeconfig files!</a></li> </ul> </li> <li><a href=""> GSoC: katacontainer support in containerd, by Jian Liu</a></li> <li><a href=""> linkerd 2.0 is GA</a> <ul> <li><a href="">Thomas Rampelberg tells you how to use it</a></li> </ul> </li> <li><a href=""> Cortex added to the CNCF sandbox</a></li> <li><a href="">Red Hat OpenShift Service Mesh, based on Istio</a></li> <li><a href=""> Microservice observability with Istio at Trulia</a></li> <li><a href=""> Contour 0.6 from Heptio</a></li> </ul> <h3 id="links-from-the-interview">Links from the interview</h3> <ul> <li><a href="">Dawn Chen on GitHub</a></li> <li><a href="">The Borg paper</a></li> <li>Process containers (later ‘cgroups’): <ul> <li><a href="">The first submission of containers to the Linux kernel</a></li> <li><a href="">Early coverage of process containers</a></li> <li><a href="">Paul Menage’s 2007 paper “Adding Generic Process Containers to the Linux Kernel”</a></li> <li>Dawn’s first job: tracking processes. Each job had its own GID - she would use <a href="">netlink connection tracking</a> to map processes and threads to GIDs, and, using <a href="">procfs</a>, figure out CPU and memory usage.</li> <li>Dawn’s second job: adjusting CPU usage using <a href="">nice</a></li> <li>Today we just use <a href="">memcg</a></li> <li><a href=""> Fake NUMA</a> - cut a machine into big chunks and assign them to groups of processes.</li> </ul> </li> <li><a href="">Linux Plumbers Conference</a> <ul> <li><a href="">Tim Hockin’s presentation at the Linux Plumbers Conference in 2011, talking about the work Dawn’s team were doing</a></li> </ul> </li> <li><a href="">lmctfy</a> - Let Me Contain That For You <ul> <li><a href="">In case you don’t get the joke</a></li> <li>It’s like <a href="">runc</a> and <a href="">containerd</a></li> </ul> </li> <li><a href="">SIG Node</a> <ul> <li>Node and lifecycle management</li> <li>Application management</li> <li>Container runtimes and <a href="">kubelet</a></li> <li><a href="">Node problem detection</a></li> <li>Resource management</li> <li>GPU & <a href="">TPU</a></li> <li>Security isolation</li> <li><a href="">gVisor</a> and Sandbox Pods</li> <li>Logging and monitoring</li> </ul> </li> <li>Was SIG Node the first SIG? <ul> <li>Tied with <a href=""> SIG API Machinery</a></li> </ul> </li> <li>How did we get to CRI? <ul> <li><a href=""> rktnetes was released with Kubernetes 1.3</a></li> <li><a href="">Hyper_ containers</a> (now <a href="">Kata Containers</a>)</li> <li><a href="">LXC and LXD</a></li> <li><a href="">kubevirt for running VMs instead of containers</a></li> <li><a href="">OCI</a></li> <li><a href=""> CRI was released with Kubernetes 1.5</a></li> <li><a href="">containerd</a> and <a href="">CRI-O</a></li> </ul> </li> <li><a href=""> Container RuntimeHandler</a>, so some pods can run with one runtime and some with another</li> </ul>