Chapter 1: Upsun and the decades of the open web
I get easily excited by new stuff. I don’t know how many trials of cool and interesting things I start every week. Getting a feel for the product, understanding the pricing, and testing out the features. But I also always think about how the people building the product consider their place in the universe. What’s their frame of mind? How do they see their future?
Well, in this article I’ll answer those questions for Upsun by telling you a bit of the story of how we think of ourselves and our new PaaS offering in the context of the changing relationship between modern organizations and their digital infrastructure.
Long story short
The short version of our story could be that we are a bunch of people with a bunch of love for the web. And maybe even more so for the “open web”, having witnessed its creation and tremendous impact on the world, and seeing it grow from a haphazard, joyous thing to what now structures much of the current economy. But this blog series isn’t about short stories, instead, I’ll share our vantage point on where we think the open web is now, where it’s going, and where we see our place in it. Immersing you in a journey through the decades of the web over a series of 3 chapters.
The main message of this series is that automated, composable infrastructure that leverages open source is the only way organizations can remain competitive and hedge their bets in times of uncertainty. Our ambition for Upsun is to be the most effective PaaS in supporting companies to do that.
Outsourcing vs. the new digital-native approach
Regardless of what type of organization you are, computers are likely to be handling many of your core business functions and most people will interact with whatever you do through a computer. This has led many companies to outsource some of those core functions, particularly in maintaining their software production capabilities and digital infrastructure. Adopting the mindset: let the professionals handle it. Outsourcing takes two major forms: using external companies, integrators, or agencies to develop and maintain code, or through the usage of Software-as-a-Service (SaaS) offerings.
While few companies in their right minds would choose to build a new data center for internal applications: the digital-native giants, on the contrary, have gone for full vertical integration. They develop in their own programming languages—think Google with Go, Apple with Swift, Microsoft with C#, Facebook with Hack, and now Amazon with Rust—on their own silicon, that runs on their own land, connected to their own fiber, sometimes even with their own power sources.
These fully vertically-integrated beasts have generated such economies of scale that they are now capable of renting out this capacity to other companies. And they can do so at a quality and cost that companies that don’t have the same level of integration cannot achieve. They are at the same time companies that offer a wide array of product offerings and Infrastructures-as-a-Service (IaaS), a platform for all other players. Interestingly part of their capacity to leverage economies of scale depends on the fact that the infrastructure code they are running is mostly open source: from Linux itself to virtualization, DevOps tooling, and observability. Through this, they are capable of mutualizing their costs because so much of the code that is not part of the vertical integration is actually open source, co-maintained by others. To a point where everything has become open. At least by name. Even things as totally opaque and closed as OpenAI.
Outsourced and locked in: Can you strike a balance with how you develop and maintain your software?
Open can mean quite a few things in our industry: open source, public APIs, and free services. Essentially everything that allows for engagement and adoption with the lowest possible barrier to entry. This is true for Infrastructure-as-a-Service (IaaS) providers as they are easy for users to get in, and if you are a startup, easy to get a considerable amount of resources for free. When you are in, you get the benefits of scale immediately and those benefits continue to expand. IaaS providers can leverage their own integrated systems to allow you to add capabilities (from their closed offerings) that would otherwise be exceedingly costly to create. Meaning users tend to stick with their providers as they can’t afford to move on to another strategy.
It seems like a damned-if-you-do and damned-if-you-don’t kind of situation: you can’t build everything by yourself and it’s cheaper to outsource. Then once you outsource it's cheaper to continue as it becomes too expensive to change. You get locked in.
A PaaS can set you free
The idea behind Upsun is that it doesn’t try to fight against the stream. We won’t be building our own data centers any time soon and we are not going to try to sell you on “open” but open with a bunch of asterisks. We don’t have a free tier. You need to pay us money to use our services. But the contract is clear: Upsun gives you the freedom to move whenever you want, with no vendor lock-in.
We don’t try to infect you, and we won’t and can’t bait and switch. If you have heavily invested in Terraform and Docker infrastructures or on Gitlab to try and build your own platform engineering practice, you may see how it might make sense to bet on someone whose long-term goals are aligned with yours.
Upsun tries, like Platform.sh, to be a general contract between developers, software, and infrastructure with very little that is specific to us. The Platform.sh PaaS allows you to describe how you build a piece of software, how you deploy it, what its infrastructure level dependencies are, how it interacts with storage, and how it interacts with the network. The cherry we added on top is a strong abstraction about how software changes. These days there are a lot of GitOps services and preview environments as a service, but none take it as far as we have.
Platform.sh allows you, on one side, to leverage underlying IaaS and their own economies of scale but we break that vendor lock-in. We allow you to automate so much of the lifecycle of your applications that it becomes reasonable to actually spend some time on developing your value-add, and only your value-add—like the giants do.
Upsun adds a new element to that promise. Upsun allows you to describe how the application scales. We kept the very hard constraints we put on the Platform.sh system from the beginning: you can still deploy arbitrarily complex apps in multiple languages, leverage multiple data backends, and do all of this with zero DevOps involvement. But now, with Upsun, we are also helping you with your applications’ dynamic behavior with much more built-in observability, an integrated global edge layer, and most importantly flexible scaling on both vertical and horizontal dimensions.
Upsun empowers development teams with the flexibility to build—and the firepower to run—diverse applications on a single, self-service PaaS. By fully managing infrastructure and security, Upsun frees every developer to easily experiment, quickly iterate, and confidently deploy applications at scale. Discover more about all of the features currently available on the Upsun PaaS on our dedicated features page available now.
We don’t expect the pendulum to swing back from using SaaS and outsourcing software development to companies betting on doing everything in-house. That doesn't make sense. But Upsun allows you to make bold bets on keeping some of your capabilities in-house. It also allows you to change the way you work with your outsourcing partners when you do. You are still in the loop. Still in control. You can still choose to internalize some. Talk to your developer.
Keeping your capability to produce just the code that you need is a huge competitive advantage. Losing it is a risk. Being a beholder to open players that may close the gates at any time is the result.
Upsun is useful, really useful when you are leveraging existing open-source technologies. Both in terms of the infrastructure level elements including the data fabric and the actual software frameworks that make us productive. It brings you the capability to have bespoke capabilities at a cost structure that is close to leveraging SaaS. For some of your core functions, I believe that is the most reasonable tradeoff. You still outsource all of the undifferentiated menial heavy lifting, but you are actually capable of developing your own capabilities—like the vertically-integrated beasts we talked about. That’s cool.
This is something that will become paramount in the coming years thanks to large language models (LLMs) and machine learning. A topic I’ll talk about in this blog series but I think before we get there, we need a bit of a historical perspective. Stay tuned for chapter 2, coming soon, where we dive right into the decades of the web we’ve witnessed so far and the technologies that have been inspiring, infuriating, and instrumental in leading us to where we are today.