Microsoft is stepping up its game around container and Kubernetes

Google open sourced Kubernetes in 2014 as a Trojan horse—bringing developers around on its way of com
Microsoft is stepping up its game around container and Kubernetes
By ARCHITECHT • Issue #119
Google open sourced Kubernetes in 2014 as a Trojan horse—bringing developers around on its way of computing before ultimately bringing them onto its cloud—but don’t be surprised to see Microsoft as the company with the most complete Kubernetes vision.
On Wednesday, Microsoft took the important step of joining the Cloud Native Computing Foundation—the open source foundation set up originally as an independent home to Kubernetes, and now to many more projects—and the seemingly smaller, but possibly more important, step of announcing a new service called Azure Container Instances (ACI). Microsoft has been doing a lot in the Kubernetes space for at least the last year (it hired Kubernetes co-creator Brendan Burns last July), which makes it joining the CNCF important mainly in the sense of solidifying that commitment. Oh, and leaving AWS as the only major cloud provider to not be a member.
What’s remarkable about ACI is that Microsoft has taken the process of building  and deploying containers—already a relatively simple process—and made it even simpler. They’re even billed by the second, which makes sense given very temporary nature of many containers. You can read more details on the specifics in the Microsoft blog post linked to above, and also here:
What I find most interesting, however, is that Azure Container Instances have nothing to do with Kubernetes out of the box. They’re just plain, old, individual Linux (and, soon, Windows) containers without an orchestrator. This could end up being a great way to get customers started experimenting with running containers and microservices on Azure, without the overhead of configuring clusters and going all Kubernetes out of the gates.
However, Microsoft has built a connector that allows ACIs to be launched and managed directly from Kubernetes. The idea is users will run long-running containers on Azure VMs, and short-lived containers on ACIs, and use Azure Container Service to manage the whole thing. 
The ACI service comes on the heels of some significant open source activity by Microsoft, too. In April, it acquired Kubernetes startup Deis, which was responsible for some important projects, such as Helm, aiming to simplify the creation of Kubernetes services. And then, in May, Microsoft (with help from its new team of Deis engineers) released Draft, an open source tool for converting existing applications into containerized ones running on Kubernetes.
It is, of course, premature to crown Microsoft the king of cloud Kubernetes when Google is still around. Google continues to work on Kubernetes and peripheral cloud-native projects, and is more than capable of innovating commercially around its existing Google Container Engine service. But Microsoft has been making up for lost time and appears determined to have its name associated with Kubernetes just like Google’s already is.
If, as all signs suggest, Kubernetes becomes a common platform on which many future applications are going to be built, I think both companies like their odds of being able to compete very strongly in that space with differentiated services. The same could probably be said of IBM, which is actively involved in Kubernetes, but runs a markedly smaller cloud operation. 
The odd man out is AWS, which would explains last week’s speculation that the world’s largest cloud provider is working on a Kubernetes service of its own. If the world is indeed heading toward Kubernetes and cloud-native, even AWS can’t risk looking like an outsider or, worse, ceding those workloads to its blood rivals.
For more on Microsoft’s ambitions here, check out my podcast interviews with Brendan Burns and Gabe Monroy:

Sponsor: DigitalOcean
Sponsor: DigitalOcean
Artificial intelligence
The new Launchpad Studio program will give accepted startups access to Google talent, as well as datasets and other tools to help them get started. This isn’t a replacement for an accelerator program as far as I can tell, but combined with Google’s new AI venture arm there is perhaps a natural path to funding, as well.
Something tells me this isn’t the only AI-based security product we’ll here about this week. PerimeterX takes the tack of trying to distinguish between actions carried out by humans and those by bots.
This is probably a stepping stone for a whole new way of doing special effects not just in VR, but in movies, as well. Obviously, AI won’t replace effects experts, but will rather be a new tool in their belts for creating scenery, etc. Here’s some other recent research, somewhat related, on how Disney is using deep learning on rendering.
Well, it won one of the contests, at least—for being able to determine what actions are happening in videos. Teams from universities and research institutes in Germany, China and Singapore won other contests, including one on answering questions from textbooks.
This is a good profile on how Fei-Fei Li helped created the ImageNet dataset and competition, which helped spawn the last several years of advances in computer vision, largely driven by deep learning models.  •  Share
This is one of several unique approaches to helping cities with the challenge of figuring out which roads need work, as well as analyzing how road design affects traffic flow and accidents.
Because people are going to program reward functions in reinforcement learning systems, and humans will also have to make the ultimate calls in situations where AI systems necessarily flag lots of false positives.  •  Share
This is a good take on how to set some limits on technology—both in our dreams of how it will change the world and how soon, but also in terms of how much we should strive to make that happen so quickly.
This seems like a step beyond what we’ve seen with Watson and other systems that claim to understand text. Microsoft says its SynNet system can take supervised learning from one domain and and use that to generate questions and answers in a new domain without labeled data.
If you enjoy the newsletter, please help spread the word via Twitter, or however else you see fit.
If you’re interested in sponsoring the newsletter and/or the ARCHITECHT Show podcast, please drop me a line.
Use Feedly? Get ARCHITECHT via RSS here:
Cloud and infrastructure
She’s apparently on the short list, and she stepped down from HPE’s board. What this would mean for HPE is a fascinating discussion, as that company really needs to embrace the future or settle into the role of servicing the past.
Packet is a company I need to learn more about, honestly. It runs a bare metal IaaS service, somewhere between DigitalOcean and AWS, GCP, etc., and is now now scaling geographically. Competing with the big boys on sheer scale isn’t a particularly great idea, but having a global footprint right-sized for your customers (assuming they need it) is definitely a good idea.
This is the first I’ve heard of WalkMe, but the company’s premise is sound, especially in a world of proliferating SaaS applications. Mastering them all without some real-time assistance is nigh impossible.  •  Share
CHaP is the latest tool Netflix built to help ensure resilience of microservices, of which it runs many. Not everyone will run microservices at Netflix scale, but this is a good reminder of how complex they can be and how many failure scenarios might exist.  •  Share
Facebook CSO Alex Stamos gave a keynote at Black Hat today, in which he implored his peers to, among other things focus on problems that are truly affecting users and consider that widespread adoption of certain practices or tools might be better than building perfect tools (or users).
When you start seeing more and more of these pile up, you understand why the project gets so much attention right now. Kubernetes is not ubiquitous, but it and the architectures it helped spawn are moving there.  •  Share
This is a fair analysis of whether openness always wins out over proprietary. Sometimes, the things with the best UX and packaging win, even if they’re not the most open.  •  Share
What more is there to say: This is yet another quantum computing company taking yet another approach to building a quantum computer. The one already in the market, D-Wave Systems, catches a lot of flack, but it also just picked up another user in Oak Ridge National Laboratory.
New ARCHITECHT Show every Thursday; new AI & Robot Show every Friday!
New ARCHITECHT Show every Thursday; new AI & Robot Show every Friday!
All things data
Laszlo Korsos is joining Citadel as a managing director and chief data officer. The broader issue, of course, is the expanding ways in which hedge funds are making use of machine learning. Experience working on webscale data systems seems like a good skillset to complement the already-strong algorithmic side of trading.  •  Share
Or so say the folks at Silicon Valley Data Science. 100 days does not seem like a long time, but it’s probably enough to identify business goals and what data will help achieve them. Of course, getting that data and building a production system, model or whatever might take a little longer.  •  Share
People get hung up on deep learning and other AI techniques for medical breakthroughs, but Spark is proving really useful, too. Here, an Australian government agency details its system for analyzing genomic datasets with millions of features. Back in 2014, I spoke with David Patterson (then at UC-Berkeley) about how Spark and big data were already helping doctors diagnose diseases. That was more about applied computational medicine, whereas this genomics work is research, but it’s good to see how open source technologies can mature and end up solving important problems in fields outside their initial targets.
Did you enjoy this issue?
The most interesting news, analysis, blog posts and research in cloud computing, artificial intelligence and software engineering. Delivered daily to your inbox. Curated by Derrick Harris. Check out the Architecht site at
Carefully curated by ARCHITECHT with Revue. If you were forwarded this newsletter and you like it, you can subscribe here. If you don't want these updates anymore, please unsubscribe here.