GUN raises more than $1.5M for its decentralized database system

GUN is an open-source decentralized database service that allows developers to build fast peer-to-peer applications that will work, even when their users are offline. The company behind the project (which should probably change its name and logo…) today announced that it has raised just over $1.5 million in a seed round led by Draper Associates. Other investors include Salesforce’s Marc Benioff through Aloha Angels, as well as Boost VC, CRCM and other angel investors.

As GUN founder Mark Nadal told me, it’s been about four years since he started working on this problem, mostly because he saw the database behind his early projects as a single point of failure. When the database goes down, most online services will die with it, after all. So the idea behind GUN is to offer a decentralized database system that offers real-time updates with eventual consistency. You can use GUN to build a peer-to-peer database or opt for a multi-master setup. In this scheme, a cloud-based server simply becomes another peer in the network (though one with more resources and reliability than a user’s browser). GUN users get tools for conflict resolution and other core features out of the box and the data is automatically distributed between peers. When users go offline, data is cached locally and then merged back into this database once they come online.

Nadal built the first prototype of GUN back in 2014, based on a mix of Firebase, MySQL, MongoDB and Cassandra. That was obviously a bit of a hack, but it gained him some traction among developers and enough momentum to carry the idea forward.

Today, the system has been used to build everything from a decentralized version of Reddit (which isn’t currently working) that can handle a few million uniques per month and a similarly decentralized YouTube clone.

Nadal also argues that his system has major speed advantages over some of the incumbents. “From our initial tests we find that for caching, our product is 28 times faster than Redis, MongoDB and others. Now we are looking for partnerships with companies pioneering technology in gaming, IoT, VR and distributed machine learning,” he said.

The Dutch Navy is already using it for some IoT services on its ships and a number of other groups are using it for their AI/ML services. Because its use cases are similar to that of many blockchain projects, Nadal is also looking at how he can target some of those developers to take a closer look at GUN.

Google and Coursera launch a new machine learning specialization

Over the last few years, Google and Coursera have regularly teamed up to launch a number of online courses for developers and IT pros. Among those was the Machine Learning Crash course, which provides developers with an introduction to machine learning. Now, building on that, the two companies are launching a machine learning specialization on Coursera. This new specialization, which consists of five courses, has an even more practical focus.

The new specialization, called  “Machine Learning with TensorFlow on Google Cloud Platform,” has students build real-world machine learning models. It takes them from setting up their environment to learning how to create and sanitize datasets to writing distributed models in TensorFlow, improving the accuracy of those models and tuning them to find the right parameters.

As Google’s Big Data and Machine Learning Tech Lead Lak Lakshmanan told me, his team heard that students and companies really liked the original machine learning course but wanted an option to dig deeper into the material. Students wanted to know not just how to build a basic model but also how to then use it in production in the cloud, for example, or how to build the data pipeline for it and figure out how to tune the parameters to get better results.

Leah Belsky, Coursera’s VP of enterprise development, echoed this and noted that this kind of specialization with a focus on practical models will make the credential more meaningful for employers.

The target audience for the specialization is somebody who wants to build new skills — and that’s pretty much every developers, especially now that machine learning is making inroads in virtually every area of tech. And since it’s almost impossible to hire machine learning experts, this course will surely be attractive to many employers who want their existing workforce to gain these skills.

As Lakshmanan noted, there are plenty of use cases for leading-edge kind of machine learning models, but what these courses focus on are more of the “day-to-day models” that can bring additional value to many existing products. Because of the focus on real-world problems, Lakshmanan also noted that the course should be useful for newly minted graduates who may be more familiar with the theories of machine learning than building products.

He also noted that only a few years ago, getting started with a course like this would have been rather cumbersome, not in the least because you need relatively powerful hardware with a dedicated GPU to work productively. Now, however, thanks to the various cloud platforms that offer GPU access or even specialized hardware like Google Cloud’s TPUs, the barrier of entry has dropped significantly.

It’s worth noting that these courses expect that you are already a somewhat competent programmer. While it has gotten much easier to start with machine learning thanks to new frameworks like TensorFlow, this is still an advanced skill. It’ll surely still be a while before we see a “get started with programming in Python by building a machine learning model” course.

Looking ahead, Lakshmanan also noted that the team is looking at a next course that would build upon the existing one, but with a focus on working with unstructured data. That’s a different class of problem with its own skill set and one that’ll allow the graduates of the first course to apply their knowledge to a whole different set of data.

The Kata Containers project launches version 1.0 of its lightweight VMs for containers

The Kata Containers project, the first non-OpenStack project hosted by the OpenStack Foundation, today launched version 1.0 of its system for running isolated container workloads. The idea behind Kata Containers, which is the result of the merger of two similar projects previously run by Intel and Hyper, is to offer developers a container-like experience with the same security and isolation features of a more traditional virtual machine.

To do this, Kata Containers implements a very lightweight virtual machine (VM) for every container. That means every container gets the same kind of hardware isolation that you would expect from a VM, but without the large overhead. But even though Kata Containers don’t fit the standard definition of a software container, they are still compatible with the Open Container Initiative specs and the container runtime interface of Kubernetes. While it’s hosted by the OpenStack Foundation, Kata Containers is meant to be platform- and architecture-agnostic.

Intel, Canonical and Red Hat have announced they are putting some financial support behind the project, and a large number of cloud vendors have announced additional support, too, including 99cloud, Google, Huawei, Mirantis, NetApp and SUSE.

With this version 1.0 release, the Kata community is signaling that the merger of the Intel and Hyper technology is complete and that the software is ready for production use.

OpenPath raises $7M to help you access your office with your phone

If you’ve ever worked in an office building, chances are somebody issued you a keycard or NFC-enabled badge to open the doors to the building. Those cards and badges do their job, but they can be both cumbersome and prone to problems. OpenPath wants to do away with all of these issues and add a new level of convenience to this whole process by replacing these access cards with the phone you already have.

Until today, OpenPath, which currently has about 20 employees, remained in stealth mode since it was founded by Edgecast co-founders Alex Kazerani (CEO)and James Segil (President), together with a number of other former Edgecast execs. The founders are putting their own money into this startup and are leading a $7 million seed round. A number of institutional investors also participated in this round, though, including Upfront Ventures, Sorenson Ventures, Bonfire Ventures, Pritzker Group Venture Capital and Fika Ventures.

Over the course of the last few years, the team developed — and patented — both the hardware and software for allowing employees to securely open doors and for security teams to manage their access. Instead of NFC, the company’s so-called SurePath Mobile technology uses Bluetooth, Wi-Fi and LTE to authenticate the user. The system integrates directly with G Suite and Office 365 so that users and IT teams don’t have to create multiple user accounts to give employees access to their spaces.

Segil argues that employees have come to expect a certain level of convenience in the workplace and while our homes are getting smarter, most offices aren’t. During our conversation ahead of today’s announcement, Kazerani also stressed that the company’s platform had to be enterprise-grade and ready to be used thousands of times a day.

The OpenPath team developed its own reader hardware, which businesses have to install at their doors. The hardware uses the same wiring as existing services, though, making it easy to replace a legacy system with this new solution.

Yubico launches an SDK that lets iOS devs add support for its NFC keys

Yubico, the company behind the increasingly popular YubiKey security keys, today announced the launch of a new SDK for iOS developers that allows them to add support for two-factor authentication over NFC with the company’s YubiKey NEO keys. With this, the company now offers solutions for all the major platforms. The first company to support this new feature, by the way, is LastPass, which already supported the Yubico one-time passwords over NFC on Android.

While it was already possible to build an integration with the YubiKey NEO once iOS 11 launched, the new SDK will make it significantly easier for more developers to support these keys over NFC.

It’s absolutely critical to have a hardware-based root of trust, like the YubiKey, to establish an approved relationship between a mobile phone and the apps we use,” said Stina Ehrensvard, Yubico’s CEO and founder, in today’s announcement. “Mobile authentication methods, like SMS or push apps, cannot be considered as trusted second factors to authenticate in a mobile app setting.”

The company argues that NFC authentication is about four times faster than getting traditional one-time passwords. Developers can use the NEO keys for giving users access to an application, as well as for step-up security to initiate actions like money transfers or password resets.

Users simply have to touch the phone with their key (assuming they have an iPhone 7 or newer and run iOS 11) and they should be good to go.

“Integrating the Yubico SDK into the LastPass iOS app was a quick and painless process, mostly because the NFC API matched almost 1:1 with the Yubico SDK API,” said Akos Putz, Principal Product Manager for LastPass in the announcement. “We’re excited to offer this new authentication method for our iOS users right out of the gate, giving them another option for adding an extra layer of security to their LastPass vault.”

Packet teams up with Platform 9 and Datera to launch its new private cloud platform

Packet, the bare metal cloud provider, today announced that it has partnered with open source hybrid cloud specialist Platform9 and storage and data management service Datera to launch a new private cloud solution for businesses that want to have greater control over their platforms. Packet argues that this new solution can save businesses up to 50 percent in cost when compared to using a public cloud solution.

“What we’re providing here is the polished experience of the public cloud, but with significantly more choice and performance,” noted Zac Smith, CEO at Packet . “By combining the strengths of market leaders like Datera and Platform9 with Packet-managed bare metal, we’re able to deliver it at a fraction of the cost of traditional public or private cloud solutions.”

Packet notes that this partnership came about after Platform9 itself migrated away from AWS. The restrictions of the public cloud model, the company argues, combines with the complexity of the public cloud billing and delivery model, lead it to look for greener pastures.

Over the course of the last few months, we’ve seen a couple of similar matchups where smaller cloud providers have teamed up to provide their own solutions to better combat the likes of AWS, Azure and Google Cloud. Just last month, Packet also partnered with Backblaze and Server Central are Backblaze’s B2 cloud storage service.

Packet notes that this new solution is available in its 18 global locations, just like its OpenStack and Kubernetes offerings.


Valimail raises $25M in additional funding for its email authentication service

Valimail helps businesses ensure that nobody can impersonate them over email. That’s not a sexy business to be in, but very much a necessary one. The company’s email authentication service, which uses standards like SPF, DKIM and DMARC, is currently in use by the likes of Yelp, Uber, Fanni Mae and WeWork. Today, the company announced that it has raised a $25 million Series B round led by Tenaya Capital, with participation from Shasta Ventures, Flybridge Capital Partners and Bloomberg Beta.

This round bring Valimail’s total amount of funding to $38.5 million, including the company’s $12 million Series A round in 2016.

“Authentication is at the root of trusted communications,” Valimail CEO and co-founder Alexander Garcia-Tobar told me. “You must be able to trust the authenticity of who/what is on the other side, or the communication is meaningless. For example, no retailer would never accept a credit card without swiping it first (either physically or virtually). What happens in the credit card world needs to happen for communications.”

The funding round is coming at an important time for email authentication. The DMARC standard is now supported by over 5 billion inboxes, according to Valimail. Over the course oft he last six month, domain owners have also published more DMARC records than in the five previous years combined. In addition, all federal agencies must implement this standard, too.

Valimail promises to take care of all the hassles of setting up support for these authentication standards.

“After attempting email authentication with other solutions, I was amazed at the level of automation Valimail provides,” said JJ Agha, VP of information security at WeWork. “It eliminates the need for two FTEs, so my staff can focus on other key priorities. I consider it a ‘set it and forget it’ solution for ensuring that our employees and executives can’t be impersonated and that our email is trusted.”



Don’t expect Ubuntu maker Canonical to IPO this year

Canonical, the company best known for its Ubuntu Linux distribution, is on a path to an IPO. That’s something Canonical founder and CEO Mark Shuttleworth has been quite open about. But don’t expect that IPO to happen this year.

“We did decide as a company — and that’s not just my decision — but we did decide that we want to have a commercial focus,” Shuttleworth told me during an interview at the OpenStack Summit in Vancouver, Canada today. “So we picked cloud and IoT as the areas to develop that. And being a public company, given that most of our customers are now global institutions, it makes for us also to be a global institution. I think it would be great for my team to be part of a public company. It would be a lot of work, but we are not shy of work.”

Unsurprisingly, Shuttleworth didn’t want to talk about the exact timeline for the IPO, though. “We will do the right thing at the right time,” he said. That right time is not this year, though. “No, there is a process that you have to go through and that takes time. We know what we need to hit in terms of revenue and growth and we’re on track.”

Getting the company on track was very much Shuttleworth’s focus over the course of the last year. That meant killing projects like the Ubuntu Phone (which Shuttleworth said was “painful,”) as well as the Unity desktop environment. Instead, the company’s focus is now squarely on helping enterprises stand up and manage their private clouds — no matter whether those run OpenStack, Kubernetes or a combination of those.

That doesn’t mean Canonical has forgotten about the desktop, though. Shuttleworth told me that the desktop team still has the same size as before. He also noted that the desktop is still a passion for him.

“We took some big risks a year ago,” he said. “We cut a bunch of stuff that people loved about us. We had to see if people were going to respond commercially.” That move is paying off now, though. During a keynote earlier today, Shuttleworth noted that Canonical is now in talks for about 200 new deployments for 2018 — up from about 40 in 2017.

While the hype around OpenStack has died down considerably over the course of the last two years, Canonical is still seeing good growth there — especially now that there are only a few major players left, including RedHat, which he name-checked a number of times during both his keynote and our conversation.

Why are things going well for Canonical when others couldn’t make a business out of OpenStack? “I believe for this community — the OpenStack community — it’s really important to deliver on the underlying promise of more cost-effective infrastructure,” he said. “You can love technology and you can have new projects and it can all be kumbaya and open source. In practice, to me, most of the stuff that we saw at OpenStack was bullshit. The stuff that really matters is computers, virtual machines, virtual disks, virtual networks. So we ruthlessly focus on delivering that and then also solving all the problems around that.”

Today, Canonical can deliver an OpenStack platform to an enterprise in two weeks — with all of the hardware and services in place. “I don’t mind being a bit controversial because we are delivering the promise of OpenStack,” he said. “The promise of OpenStack wasn’t delivering endless summits and endless new projects and endless new ideas.” That, he said, is exactly the kind of bullshit he was referring to in his earlier comments.

Looking ahead, Shuttleworth noted that he’s especially interested in what Canonical can do around IoT solutions, too. Thanks to Ubuntu Core and its Snap system, it has all the tools in place, including a lightweight management layer. The company also is focusing heavily on getting more customers in the financial services sector. No doubt, having a bunch of large banks and brokerages as reference customers will help the company when it comes to its IPO — and my guess is that we can expect that one to happen next year.

OpenStack spins out its Zuul open source CI/CD platform

There are few open source projects as complex as OpenStack, which essentially provides large companies with all the tools to run the equivalent of the core AWS services in their own data centers. To build OpenStack’s various systems the team also had to develop some of its own devops tools, and in 2012, that meant developing Zuul, an open source continuous integration and delivery (CI/CD) platform. Now, with the release of Zuul v3, the team has decided to decouple Zuul from OpenStack and to run it as an independent project. It’s not quite leaving the OpenStack ecosystem, though, since it will still be hosted by the OpenStack Foundation.

Now all of that may seem a bit complicated, but at this point, the OpenStack Foundation is simply the home of OpenStack and other related infrastructure projects. The first one of those was obviously OpenStack itself, followed by the Kata Containers project late last year. Zuul is simply the third of these projects.

The general concept behind Zuul is to provide developers with a system for automatically merging, building and testing new changes to a project. It’s extensible and supports a number of different development platforms, including GitHub and the Gerrit code review and project management tool.

Current contributors include BMW, GitHub, GoDaddy, Huawei, Red Hat and SUSE. “The wide adoption of CI/CD in our software projects is the foundation to deliver high-quality software in time by automating every integral part of the development cycle from simple commit checks to full release processes,” said BMW software engineer Tobias Henkel. “Our CI/CD development team at BMW is proud to be part of the Zuul community and will continue to be active contributors of the Zuul OSS project.”

The spin-off of Zuul comes at an interesting time in the CI/CD community, which is currently spoiled for choice. With Spinnaker, Google and Netflix are betting on an open source CD platform that solves some of the same problems as Zuul, for example, while Jenkins and similar projects continue to go strong, too. The Zuul project notes that its focus is more strongly on multi-repo gating, which makes it ideal handling very large and complex projects. A number of representatives of all of these open source projects are actually meeting at the OpenDev conference in Vancouver, Canada that’s running in parallel with the semi-annual OpenStack Summit there and my guess is that we’ll hear quite a bit more about all of these projects in the coming days and weeks.




Nvidia’s researchers teach a robot to perform simple tasks by observing a human

Industrial robots are typically all about repeating a well-defined task over and over again. Usually, that means performing those tasks a safe distance away from the fragile humans that programmed them. More and more, however, researchers are now thinking about how robots and humans can work in close proximity to humans and even learn from them. In part, that’s what Nvidia’s new robotics lab in Seattle focuses on and the company’s research team today presented some of its most recent work around teaching robots by observing humans at the International Conference on Robotics and Automation (ICRA), in Brisbane, Australia.

Nvidia’s director of robotics research Dieter Fox.

As Dieter Fox, the senior director of robotics research at Nvidia (and a professor at the University of Washington), told me, the team wants to enable this next generation of robots that can safely work in close proximity to humans. But to do that, those robots need to be able to detect people, tracker their activities and learn how they can help people. That may be in small-scale industrial setting or in somebody’s home.

While it’s possible to train an algorithm to successfully play a video game by rote repetition and teaching it to learn from its mistakes, Fox argues that the decision space for training robots that way is far too large to do this efficiently. Instead, a team of Nvidia researchers led by Stan Birchfield and Jonathan Tremblay, developed a system that allows them to teach a robot to perform new tasks by simply observing a human.

The tasks in this example are pretty straightforward and involve nothing more than stacking a few colored cubes. But it’s also an important step in this overall journey to enable us to quickly teach a robot new tasks.

The researchers first trained a sequence of neural networks to detect objects, infer the relationship between them and then generate a program to repeat the steps it witnessed the human perform. The researchers say this new system allowed them to train their robot to perform this stacking task with a single demonstration in the real world.

One nifty aspect of this system is that it generates a human-readable description of the steps it’s performing. That way, it’s easier for the researchers to figure out what happened when things go wrong.

Nvidia’s Stan Birchfield tells me that the team aimed to make training the robot easy for a non-expert — and few things are easier to do than to demonstrate a basic task like stacking blocks. In the example the team presented in Brisbane, a camera watches the scene and the human simply walks up, picks up the blocks and stacks them. Then the robot repeats the task. Sounds easy enough, but it’s a massively difficult task for a robot.

To train the core models, the team mostly used synthetic data from a simulated environment. As both Birchfield and Fox stressed, it’s these simulations that allow for quickly training robots. Training in the real world would take far longer, after all, and can also be more far more dangerous. And for most of these tasks, there is no labeled training data available to begin with.

“We think using simulation is a powerful paradigm going forward to train robots do things that weren’t possible before,” Birchfield noted. Fox echoed this and noted that this need for simulations is one of the reasons why Nvidia thinks that its hardware and software is ideally suited for this kind of research. There is a very strong visual aspect to this training process, after all, and Nvidia’s background in graphics hardware surely helps.

Fox admitted that there’s still a lot of research left to do be done here (most of the simulations aren’t photorealistic yet, after all), but that the core foundations for this are now in place.

Going forward, the team plans to expand the range of tasks that the robots can learn and the vocabulary necessary to describe those tasks.