It’s the beginning of the year, and, as a result, everybody is offering their predictions for 2019. View in browser »
The New Stack Update

ISSUE 148: Kubernetes, the Virtual Machine Orchestrator?

Talk Talk Talk

“Everyone wants actual isolation, but to be honest I’m not sure Kubernetes is the best abstraction for the job. It’s so complex and leaves so much room for error. One config gone wrong or one missed patch and it’s game over for your cluster.”

GitHub’s Jessie Frazelle
Add It Up
Commits in Public Repositories
Recent reporting claims computer science students don’t want to work for Facebook. Based on your ethics as a developer, would work for Facebook? If open source is your thing, then perhaps you will bring some love to the recent Facebook hate fest.

For the privacy-minded, this past year has offered plenty of reasons for avoiding the social media giant, most notably its murky, ever-shifting privacy policies and seeming eagerness to share user data with other companies, from Cambridge Analytics to Netflix.

But Facebook is making significant contributions to leading-edge, widely used open source projects. GitHub's annual Octoverse report says Facebook is fourth in the world in terms of corporations contributing to major repositories. Our own reporting showed that Facebook was in sixth place, but still punching above its weight class. The company has 474 open active repositories in its portfolio. Impressively, 30 percent of the 94 thousand commits to these projects last year came from people that do not work at Facebook. Standouts among Facebook’s project include React, which has over 10,000 contributors, and Pytorch, a machine learning platform that happens to be the second fastest growing major project according to the GitHub report.
What's Happening

According to those who responded to The New Stack's reader survey — working mostly in development and/or DevOps or operations — the trends and topics you are especially interested in include cloud native artificial intelligence (AI) and machine learning (ML) on Kubernetes and serverless. DevOps, as well as security, of course, also play a big role, as data is now being processed, managed and stored in new and exciting ways.

It was with these topics in mind that Alex Williams, founder and editor-in-chief of The New Stack, along with Joab Jackson, TNS managing editor, hosted the last TNS podcast of 2018. The guests were Dillon Erb, CEO of Paperspace, which offers solutions for AI and ML deployments on the cloud, and Chenxi Wang, managing director of venture capital firm Rain Capital, whose emphasis is on next-generation security solutions.

The New Stack Survey: What to Expect in 2019

Kubernetes the Virtual Machine Orchestrator?

It’s the beginning of the year, and, as a result, everybody is offering their predictions for 2019. We take most prognostics with a grain of salt, though Pivotal principal technologist Paul Czarkowski caught our eye with a post predicting that 2019 will be the year of Kubernetes … for virtual machines. Not containers, but virtual machines.

“The future of Kubernetes is Virtual Machines, not Containers.” Czarkowski proclaimed in a blog post.

The problem, as Czarkowski points out, is that it is difficult to secure containers in multitenant environments, given they all share the same kernel. “Linux containers were not built to be secure isolated sandboxes (like Solaris Zones or FreeBSD Jails). Instead, they’re built upon a shared kernel model that utilizes kernel features to provide basic process isolation,” he wrote. In particular, Czarkowski looked at the feasibility of setting up of “hard multitenancy, in which multiple tenants of a cluster should assume all the other tenants are inherently untrustworthy. Avoiding hard multitenancy, organizations have been instead setting up lots of small Kubernetes clusters that run independently, creating what Czarkowski calls “Kubesprawl.”

While Kubernetes has a multitenancy working group to tackle this challenge of providing full isolation for containers, Czarkowski suggests another solution, namely virtual machines. VMs, with their own OS kernels, are inherently multitenant ready (after a bit of configuration). And we are already seeing a lot of work along these lines, he points out. The open source Virtual Kubelet project, for example, was designed as a way to manage containerized virtual machines. Other secured-container approaches built on slimmed-down virtual machines are also emerging, including Kata Containers, Amazon’s Firecracker, and Google’s gVisor.

So, virtual machines, the hot technology of 2007, may also be a top technology of 2019?

Mind-Reading, Self-Replicating: A Look Back at the Year’s Best AI Stories

It was quite a year for the field of artificial intelligence. Our AI science reporter Kimberly Mok recounts a few of the most notable advances in 2018, including an algorithm to decode brain waves to attempt to read a person’s thoughts, machines that can predict future actions using “visual foresight,” and the development of mathematics to mimic social skills, or the cooperation across machines acting in the same manner as people do in a group project.

Tech Ethics New Year’s Resolution: Don’t Build Software You Will Regret

Developers who may want to take the new year as an opportunity to link their personal ethics with the work they do for a living may want to read Jennifer Riggins’ post about ethics as software development. With ethical software, the idea is to cause no negative social impact, and to not make the world a worse place to live in.

Linux Technology for the New Year: eBPF

In the year to come, we will start to see a change in the Linux kernel architecture as a new component, eBPF, starts taking over more monitoring, security and networking duties from individual kernel modules.

Party On


Congratulations to the winner of The New Stack reader survey sweepstakes, Martin Hueser! Martin is a software engineer and architect, living and working in Germany. At age 50 he has over 20 years of experience, mostly with telecommunications-related software for small and medium enterprises. 

Free Serverless Ebook

Experts and visionaries in distributed systems believe serverless technologies are the next evolution of application infrastructure beyond microservices. Leading edge companies have already embraced serverless and have dramatically reduced operational overhead and streamlined the DevOps cycle, while increasing scalability and resiliency. Still, there are many challenges to serverless adoption, such as operational control, complexity and monitoring.

The New Stack’s Guide to Serverless Technologies will help practitioners and business managers place these pros and cons into perspective by providing original research, context and insight around this quickly evolving technology. 

Download The Ebook
We are grateful for the support of our ebook sponsors:

Copyright © 2019 The New Stack, All rights reserved.

Want to change how you receive these emails?
You can update your preferences or unsubscribe from this list