Cloud Native ComputingDevelopersFeaturedSuccess StoriesVideo

Why Jarvus Chose Linode over AWS, GCP and Azure?


“When we put clients on top of Kubernetes offering from AWS, Azure or GCP and something goes wrong, you would try to get support from one of those three companies, you’d talk to someone on the other side of the world whose job is to get you away as fast as possible. Talking to someone at Linode is a whole other experience. We know they’re here in Philadelphia and they actually try to solve your problem, which sounds silly to be the difference, but it really is a huge difference. We can trust Linode to take care of our clients” – Chris Alfano, CTO,  Jarvus.


Here is the lightly edited transcript of the interview:

Swapnil Bhartiya: Hi, welcome to TFIR Success Stories. Today we have with us, Chris Alfano, CTO of Jarvis. Chris, first of all, welcome to the show. Tell us a bit about the company Jarvis.

Chris Alfano: Jarvis helps mission-driven communities and organizations craft their own open source cloud technology and put it to work for people on the ground. When we first started Jarvis, I was working full time inside a Philly public school, and a lot of our early work was focused on working with schools and non-profits and community events and small organizations around Philadelphia. As we grew into helping some of the world’s largest enterprises, we tried to maintain this balance of ground local work and this Skyhigh enterprise work. Something that always struck us was how there was this whole world of amazing services of innovation being dedicated to empowering enterprises in the largest companies to make use of technology, but at the same time, smaller organizations just weren’t having things designed for them. Too often, they were ending up almost as sharecroppers, using software that they don’t own, that they can’t change.

What’s ironic about this is so much of this innovation in the enterprise space is built on the backs of open source software. Things like Linux and Git and my MySQL, that were built with these strong copyleft licenses and these ideas of empowering people to not just use the software, but be able to change the software they use, be able to use it as they see fit, being able to share it with their peers. We learned a lot along the way, and today Jarvis is shifting gears to work on bringing what we’ve learned to the masses. By masses I mean, mission-driven communities that work together, work on health, learning and play. The things that make up our everyday lives.

Swapnil Bhartiya: Can you give some examples of what kind of people or organizations do you help?

Chris Alfano: Building on my work initially in schools, when we first started Jarvis, we very quickly started partnering with innovative schools around the world. The school I worked with was called The Science Leadership Academy here in Philadelphia. It was started by a guy named Chris Lehmann that I saw speaking in a dark basement in Philly at a night talk. He talked about how schools needed to be reinvented with technology, how what we had right now is schools plus technology. It’d be like, if when the printing press was invented, you just had Europe plus the printing press, but really what you got was a whole new Europe. The same transformation needs to happen with schools. He talked about how schools needed to be inquiry-driven, they needed to focus on producing citizens, not workers. This all really appealed to me and my collaborators at the time.

We had wanted to build software for schools, but we had been thinking about it in the traditional Silicon Valley approach of, we’re going to build something that’s going to grow quickly, we’re going to skip working with administrators and districts in schools because they’re too slow, it’s not going to grow fast enough. We’re going to be really top-down and revolutionize and disrupt. We took this sharp turn when we partnered with the school, and we found that all over the world, there are these communities of experts who aren’t software developers, they’re educators, they’re starting schools, they’re teachers, and they’re just trying to make do with the software they have. They’re hacking it, they’re getting things done. By and large, the technology community isn’t helping them.

We’re not building things to serve their innovation. We’re trying to build things and then grow them nationally and ship them to every school in the country, but we’re not respecting the learnings on the ground and giving people the tools to iterate. As we sought schools to partner with, we built this common foundation and we started working with one, two, and then three different groups of schools that were all pursuing different visions of how schools could be a re-imagined with technology. We quickly learned that there’s no one right way, and that really, the right way is distributing the ability to create software and new software.

One concrete example of that here in Philadelphia, there was a school called Building 21. We partnered with them to implement a competency-based learning grade book, which is totally different than any grade books online. They had a very new learning model, software companies weren’t keeping up with them, so we helped them build that software and work on it year after year in response to what they learned as they actually put it in front of students.

Swapnil Bhartiya: What kind of services or products are you offering? I want to understand the technology stack that you make available to these users of yours.

Chris Alfano: Primarily what Jarvis has done until now has been case-by-case consulting. We’ve really functioned as management consultants that also develop software, and we haven’t really put major investment into building platforms and products. It’s something we’ve done as needed, but that’s part of the shift we’re looking to make over the next decade, to start taking the lessons we’ve learned and bottling them up a little more for other people to follow in our customer’s footsteps.

Swapnil Bhartiya: What kind of infrastructure are you, yourself using to build these applications or tools for your users?

Chris Alfano: Early on we started with just renting servers and boxes all over the world and then running them ourselves. When we first got started, there wasn’t anything like Kubernetes, so we had to keep track of virtual machines and hard drives. We had to focus on all this low level stuff. When the cloud revolution started, and virtual machines became things you could rent all over the place, we still stuck to manually managing it because we didn’t see any real comprehensive way to do it better. Since Kubernetes came on the scene, we’ve been watching it cautiously for the first couple of years, we saw enterprises adopt it in mass, we saw the major three cloud providers adopted it, and those were really encouraging signs for us, but it still seemed like it was hard to access for the customers we cared about, the smaller organizations.

The big three providers, they’re really focused on helping companies solve problems by throwing money at it. It just wound up being way too expensive for the small schools and organizations we work with. It wasn’t until Linode came along with their Kubernetes offering, that we finally started seeing Kubernetes be viable for the scale of clients we serve.

Swapnil Bhartiya: Why and when you decided to switch to Linode? What unique value did you see in Linode?

Chris Alfano: We’ve been fans of Linode for a long time because they’re a Philadelphia company and we’re a Philadelphia company, so there’s that obviously. But it really didn’t make sense to pitch to our clients until the Linode Kubernetes offering came along. We were early adopters in their beta program and had a lot of success using their early versions of the tool. We found it very advantageous because of their simple pricing, flat fees. Our clients need simple, reliable service, and they need to be able to call someone when something goes wrong.

When we would put our clients on top of GCP’s Kubernetes offering, or Azure or Amazon’s, when something would go wrong and you try getting support from one of those three companies, you’d talk to someone on the other side of the world who, it’s clearly their job to get you away as fast as possible. Talking to someone at Linode is a whole nother experience. We know they’re here in Philadelphia and they actually try to solve your problem, which sounds silly to be the difference, but it really is a huge difference. We can trust Linode to take care of our clients.

Swapnil Bhartiya: From the perspective of services, how different is Linode’s offering vis-a-vis, as you said, the big three players, which is GCP, Azure, and AWS?

Chris Alfano: I think the big three providers focus a lot of energy and time and your money on solving problems of rapid unmitigated scale. The kind of problems you have when you’re a global company, you’re trying to solve everything globally. That’s just not the customer we work with. There’s nothing wrong with that way to approach problems, but there’s a whole segment of the economy where folks are limited to what we call human scale. If you’re a school or a city agency or a community event, you’re interacting with real humans, you’re inherently unscalable. The way that those organizations and operations scale is that there are thousands of them around the country. Not one that grows up to be enormous.

You’re talking about very different challenges. They need economic and predictable tools, they don’t need to worry about a global scale. They don’t need all the things that make the big three providers offerings complex and expensive, they need something simple and predictable, and that’s what Linode provides. Linode’s Kubernetes offering is just tuned to a different market, different audience.

Swapnil Bhartiya: How are you leveraging LKE or Linode’s Kubernetes in general?

Chris Alfano: A big part of Jarvis’ innovation’s pitch to our customers is the notion of software freedom. When the open source revolution first kicked off, it was all about tools that you ran on your computer, and the folks that built the underpinnings of all the technology we use today, they adopted these copyleft licenses, which says, if you give someone software, you also have to give with them the ability to change that software and use that software. This is very important. If you’re building a business or an organization, you’re investing your time, your investor’s money, your backer’s money, your employee’s time, your volunteer’s time, you’re building something, you want it to last. What happens when you build that on top of proprietary SaaS offerings, things from companies that might change their business model next year, that might get acquired and be gone?

You really can’t fully invest in technology if you can’t trust that you can change it as your organization evolves, or keep using it as your economics change, or keep using it as the ownership of that company changes. The customers that Jarvis serves, they’re very concerned about that notion of what does their future look like five years from now, 10 years from now, if they’ve set up their entire business on the back of someone’s SaaS startup, which could disappear overnight or change the business models overnight? We talked about how important it is to build on top of open source software and have all of that freedom.

Now, something that happened when we shifted from the personal computer revolution to the cloud revolution, was real people lost the agency to make use of free software. Free software used to be something you downloaded on your computer and you could use it, and then if you wanted to change it, you could learn the right code, that’s one option. You could hire someone or find a friend to write code, but I think more importantly, given my example of a school trying to implement new learning models, what we saw out on the ground was that not every principal was a coder, but in a group of five, 10, 20 principals who were trying to start new schools, one of them knows a coder, one of them starts hacking. They invent new configurations, they invent new software and they share it with their peers. Even if you’re not a coder, you can gain the benefit of other people’s innovation when they build on top of open source software.

We lost that in the transition to the cloud, because now, to deploy pieces of open source software, you need infrastructure, you need routing, you need networks, you need all of these components, you can’t just get a computer from the store, take it out of a box, put it on your desk and be up and running. But Kubernetes, as a technology, finally brings that vision, that possibility into reality. Kubernetes on its own, its technology doesn’t quite do that, but the ubiquity it’s attained across all the service providers, and now with Linode and others offering a service scale to the small organization, now we’re finally in a place where it is really possible for a small organization, for a small business to build a technology stack that they really are empowered to operate and run themselves.

Just with open source software, if you hire a firm to set something up for you, but you can’t operate it yourself, you’re still basically using proprietary software because only one vendor can change it for you. But with Kubernetes, and the ability to port your workload between vendors like Linode, we can finally give our client’s software and they actually own it. We set them up on a Linode account, we set them up with their own GitHub account, and then we just build their software and they pay us as long as we’re adding value to their business. But they’re not stuck with us. They’re truly using open software because they truly can change it themselves, keep using it without us, hire someone else to replace us or work with us to respond to their evolving needs.

Swapnil Bhartiya: What role does Jarvis play or Linode play, in making these open source technologies more accessible to a wider set of users without them having to invest their resources into just managing those open source technologies?

Chris Alfano: That’s a great question, and that really is what’s holding back this dream of ubiquitous open source software today. Jarvis believes that really, there’s a big opportunity right now, just in shifting priorities. Up until now, all the best practices and software development, all the tools and frameworks and infrastructure that we build, are oriented around this Silicon Valley dream of building something in your garage, and then overnight it explodes and you have to grow fast and you have to capture a lot of users and you have to get someone to want to come and buy the users that you’ve captured. We think Kubernetes is just the first building block of a whole new world that’s now possible. We think that there’s an opportunity to really radically shake up different parts of the software stack.

When you forget about needing to support a global scale, that that space is well covered. Lots of companies, lots of developers, lots of tools are built around solving global-scale problems. But what about the thousands of schools around the world and cities and small organizations and non-profits and communities? They all need to build software too, and they don’t need to worry about serving millions of users overnight. That’s not a problem that’s in their domain. There’s a lot of advantages we think you can find when you take the foundation of a predictably priced, simple Kubernetes offering that Linode gives, that can give a small business for $20, $40, $60, $80 a month, a solid foundation of their computing storage needs, and then everything else is defined by software. The software can be innovated, the software can be broken into tiers and maintained communally.

We think there’s an opportunity now to develop a layer of software on top of Kubernetes, that is really focused on a small scale, not application development, but environment development. I’ll give you one example. Something very popular in the enterprise is this notion of microservices. Building little tiny components for every single team. It’s a way to model your software after your enterprise so that you can avoid these problems of all your different teams coordinating together. Well, smaller organizations don’t have that problem. When you try to put a microservice architecture in front of them, it’s a ton of complexity, it’s a ton of moving pieces. A lot of software developers will look at a school and ask the question, could they run their own software and think about all these enterprise tools and assess, well, it doesn’t make any sense. Well, of course, it doesn’t make sense. Those tools weren’t designed for their scale. They were designed for a very different, much more complex problem.

The type of scale we really need to focus on in-ground software, as opposed to cloud software, is not how do you reach one instance that serves millions of people, but how do you have one instance that 10 people can copy and then one person can iterate all a little bit and share it with 10 other people? It’s a very different type of scale that maximizes human creativity on the ground. The creativity and problem solving of teachers and principals and folks that are serving communities, person-to-person. If you start from there, I think you arrive at very different designs for a lot of pieces of the software stack. I think there’s a lot of opportunities there. We don’t know all the answers yet, but hopefully, we’ve had a head start on working on them.

Swapnil Bhartiya: Chris, thank you so much for taking your time out and explaining what Jarvis does and how you leverage… Not only you are actually leveraging a lot of these open source technologies, which companies like Linode are making easier to consume, you’re actually also promoting this usage of these technologies so others can also leverage, and they can also build the tools and services they need for their own purposes in this example. Schools are a very good example because they are strapped resources sometimes, and open source does cost a lot. You don’t have to pay those heavy licenses. Thanks for explaining all that, and I look forward to talking to you again. Thank you.

Chris Alfano: Thank you, Swap.