Radio Cloud Native – Week of May 5th, 2022

Nick Chase & Eric Gregory - May 6, 2022 - , ,

Every Wednesday, Nick Chase and Eric Gregory from Mirantis go over the week’s cloud native and industry news.

This week they discussed:

You can watch the full replay below:

To join Nick and Eric next Wednesday, May 11, at 1:00pm EST/10:00am PST, register here.

Kubernetes reaches 1.24

Eric Gregory: Yesterday was 1.24 day for Kubernetes, after a short delay last month. The release is codenamed Stargazer, and the big story is the deprecation of dockershim, which we’ve talked about at some length over the last couple of months. Kubernetes users who are relying on Docker Engine as their container runtime within the cluster will need to either use the cri-dockerd adapter or shift to a runtime such as containerd or CRI-O. This is a breaking change, so cluster administrators will want to make sure they have their ducks in a row before upgrading.

We’ve talked a lot about the dockershim deprecation already, so what else is going on in version 1.24? First off, new beta APIs are now turned off by default. This only applies to new beta APIs, so existing ones will still be available by default, but it’s a significant policy change that is worth flagging.

There’s also a quiet theme in this release, improving the way the system handles volumes.

An alpha feature enables Container Storage Interface drivers to monitor volume health with a sidecar. Meanwhile, resizing persistent volumes with the CSI is now finally considered a stable feature. Also graduating to stable is a feature that prevents pods from being scheduled to a node where CSI volumes don’t have enough space to accommodate the pod’s needs, even if all the other node resources are aligned. All in all, these changes should make CSI activity more predictable and observable.

Finally, 1.24 introduces an alpha feature for signing release artifacts and verifying the components of Kubernetes itself, measures aimed at enhancing software supply chain security. The real change here is trying to introduce a standardized strategy for doing this, so I think the notable point here is that software supply chain security is increasingly becoming a priority.

Source: Kubernetes 1.24 Release Information

CNCF considers working group to examine environmental impact

Nick Chase: The CNCF Technical Oversight Committee is considering a new working group to cover Environmental Conservation/Sustainability. The group’s charter points out that data centers currently use 2% of the world’s energy and that that is expected to grow to 12% by 2040. The group will work to not only reduce use of greenhouse gases in computing, but “to raise awareness of environmental sustainability as a key element of open source development and support projects which foster an understanding of energy drivers.”

Surveys on the state of the cloud marketplace

Eric Gregory: Recent surveys gives us some insight on the cloud big three: AWS, Azure, and Google Cloud. According to the Synergy Research Group’s analysis of Q1 earnings data, those three collectively hoovered up 65% of global cloud spend. The cloud market just keeps growing, with a year on year increase from 2021 is 34%.

Meanwhile, earnings reports also indicate that colocation providers like Equinix are showing strong growth; Equinix was up six percent year on year in spite of supply chain issues and inflation. On the earnings call, CEO Charles Meyers said that Equinix has, quote, “43 projects underway across 29 metros in 20 countries, including new projects in Atlanta, Mumbai, Sydney, Tokyo, and Washington DC.” This doesn’t factor in the company’s recent acquisition of West African infrastructure provider MainOne (and its subsea cable system) or the planned acquisition of Chilean data centers next quarter.

Sources: The Register on the Big Three and colocation providers.

Algorithmic bias

Nick Chase: Last week we talked about EU regulations that will start to regulate the explainability of Artificial Intelligence algorithms in use by large companies, and this week we’re talking about new regulations that will actually talk about how they can be used. For example, in 2023 New York City will start restricting how companies can use AI tools for employment decisions, both for recruiting and promotion recommendations. The law won’t only require companies to notify applicants that they’re being evaluated by an AI and give them the option to request an alternate form of evaluation, but it will also require that companies submit their AIs to a so-called “bias audit,” in which an independent auditor checks the algorithms to ensure that they’re treating all applicants fairly.

And this is actually more important than you might realize, because it turns out that artificial intelligence is in use more than you might realize. Five years ago PriceWaterhouseCoopers released a study that showed that 40% of international companies were using artificial intelligence in Human Resources decisions, which referred not just to analyzing keywords on resumes, which of course we all knew was going on, but also, apparently, analyzing video interviews.

When Illinois became the first state to regulate AI in employment decisions, the state legislator who introduced the bill, Jaime M. Andrade, Jr., said “I found that most people think that three-minute video they send in is just to an interviewer, but what they didn’t know is they were being analyzed by a pre-screener system.”

And all of this is important to keep in mind because bias is extremely difficult to completely remove from a system. For example this week The Register reported on a paper published in the journal Science that showed that just training a model with enough data still might not be enough to remove bias. Researchers started with a functional MRI, or fMRI model that was trained 50/50 with white and African-Americans, and saw much lower accuracy for African-Americans. Then they trained the model only with African-Americans and accuracy for African-Americans went up — but it was still lower than for white Americans.

And all of this boils down to how you build out the model in the first place. If you have a model that decides college admissions by zip code instead of grades, no amount of training is ever going to make it fair. And sometimes, as in the case of the fMRI study, there are just factors you’re not aware of. For example, Jingwei Li, a postdoctoral research fellow at the Institute of Neuroscience and Medicine, Brain and Behaviour from the Jülich Research Centre in Germany, told The Register, “Several steps during neuroimaging preprocessing could have influenced the result. For example, during preprocessing, a convention is to align individuals’ brains to a standard brain template so that individual brains can be comparable. But these brain templates were usually created from the White population.”

So we don’t always have control over what the model does so it’s important to at least know where the biases are. For example, the New York law requires companies to do a bias audit and post it on their website, but doesn’t actually require the elimination of bias.

So if you’re a company and you see these regulations coming your way, what do you do? Well, TechRepublic recommends the following:

  • The first and most important thing is to at least try to start with a large and diverse set of data so you can head off the most obvious cause of bias, which is limited data sets.
  • Once you’ve done that, you should have someone who actually understands how to do a bias audit take a look at the system and keep testing it until you’ve weeded out as much buried bias as possible.
  • And then finally, simply don’t let the system exclude candidates. Have it provide information and guidance, but decisions should still be made by humans.
From Virtualization to Containerization
Learn how to move from monolithic to microservices in this free eBook
Download Now
Radio Cloud Native – Week of May 11th, 2022

Every Wednesday, Nick Chase and Eric Gregory from Mirantis go over the week’s cloud native and industry news. This week they discussed: Docker Extensions Artificial Intelligence shows signs that it's reaching the common person Google Cloud TPU VMs reach general availability Google buys MobileX, folds into Google Cloud NIST changes Palantir is back, and it's got a Blanket Purchase Agreement at the Department of Health and Human …

Radio Cloud Native – Week of May 11th, 2022
Where do Ubuntu 20.04, OpenSearch, Tungsten Fabric, and more all come together? In the latest Mirantis Container Cloud releases!

In the last several weeks we have released two updates to Mirantis Container Cloud - versions 2.16 and 2.17, which bring a number of important changes and enhancements. These are focused on both keeping key components up to date to provide the latest functionality and security fixes, and also delivering new functionalities for our customers to take advantage of in …

Where do Ubuntu 20.04, OpenSearch, Tungsten Fabric, and more all come together? In the latest Mirantis Container Cloud releases!
Monitoring Kubernetes costs using Kubecost and Mirantis Kubernetes Engine [Transcript]

Cloud environments & Kubernetes are becoming more and more expensive to operate and manage. In this demo-rich workshop, Mirantis and Kubecost demonstrate how to deploy Kubecost as a Helm chart on top of Mirantis Kubernetes Engine. Lens users will be able to visualize their Kubernetes spend directly in the Lens desktop application, allowing users to view spend and costs efficiently …

Monitoring Kubernetes costs using Kubecost and Mirantis Kubernetes Engine [Transcript]
Service Mesh for Mere Mortals
A Guide to Istio and How to Use Service Mesh Platforms
Technical training
Learn Kubernetes & OpenStack from Deployment Experts
Prep for certification!
View schedule
Mirantis Webstore
Purchase Kubernetes support