OpsCanvas: Infrastructure as Code for Smaller Shops
With the rise of Infrastructure as Code, too often the tools in the market are focused on the enterprise and large teams, while startups and other small businesses are left struggling.
That’s the opportunity Brian Kathman and Jason Turim see with their latest venture OpsCanvas, which just came out of stealth and is generally available. Its “draw and deploy” design aims to eliminate the learning curve with IaC and just get applications deployed to the cloud quickly.
“A lot of those solutions aren’t playing at [small shops’] doorsteps, so it requires them to either over-invest in these power tools — some of the larger packages out there that take weeks and months to get to Day 1 — or they have to deploy it manually.
“And so we feel like the opportunity here for us is to transform the game [with a] subscription-based SaaS deployment platform that will eliminate a lot of these hurdles of investing in platform engineering, and really out of the box, provide them with key operations that automate the process,” said OpsCanvas CEO Kathman.
Aimed at Small and Medium Businesses
Kathman and Turim previously created Signal Vine, a text messaging platform aimed at universities. There they encountered the problems first-hand. Too often small shops don’t have the budgets to really invest in IaC or to hire and retain DevOps professionals. At the same time, deploying IaC in the cloud tends to be too manual, requiring expertise to get started and keep applications in top shape.
“This pain was acute for us,” said Turim, OpsCanvas CTO. “We had one DevOps person, and basically everything went through him. And it was difficult for him, because vacations and what have you were always sort of challenging for the team. There was like a laundry list of stuff that he wanted to work on, but keeping the lights on and getting software out the door was always top of his list. We had all the pipelines and all the sort of standard, what you think of as like IaC starter stuff implemented, but it’s rarely just as smooth as that, has been our experience.”
Deploy from a Diagram
Their solution to this is to take a visual approach. Using what they call Generative IaC, the platform analyzes application architecture diagrams and automatically generates IaC code — it can create more than 10,000 lines of custom configurations in less than 30 seconds.
It’s one-click automation for common tasks involved with deploying software to the cloud, such as the configuration and provisioning of managed services including networking, database and other cloud services, and launching containerized applications to take advantage of the appropriate level of cloud resources. It enables teams to automate cloning, updating, rolling back, extending and decommissioning resources.
It takes an opinionated view of the best approach to deployment, though users can override its configurations. For instance, it uses T-shirt sizing — small, medium, large — and a user could opt for a bigger database, for example. Those configurations are not a black box; users can see everything if they want to or not look at any of it if they don’t.
You can see everything in one place including environments, versions and more.
OpsCanvas generates standardized, well-tested configuration code incorporating best practices, which the company touts as more secure by eliminating the risk of human error that can leave security holes and can lead to long debugging times.
It automates deployment to Amazon Web Services (AWS) so far, but plans are in the works to expand to Azure and Google Cloud Platform as well.
No Expertise Required
The problem with big players in IoC like Terraform or Pulumi, according to Turim, is they require learning a programming language that supports the SDK and rethinking the infrastructure in that vein. Users have to set up governance and learn the stack, then wire that into their GitOps pipeline.
“Our whole proposition is that this is orthogonal to the business that you’re trying to do, which is to build software,” he said.
“The way we’re different is we’re saying, ‘You just don’t have to think about that, you just focus on building your software, give us Docker containers, which you probably have anyway, and we can deploy. So draw this diagram that we know you have in your head, and we can deploy from that, instead of having to build out processes and learnings.’”
“We’re just saying, ‘Don’t even worry about any of that because we put all of that under the hood for you.’ We run GitOps, we’re running Terraform, but you as a developer and development team, just think about it in terms of your application, the components you write and the components in the cloud that you use.”
The application can be in any programming language, but it must be containerized to use OpsCanvas.
Kathman pointed to three differentiators:
- It’s an end-to-end deployment platform, deploying the network, managed services and the application all at once, rather than solutions more focused on just network infrastructure configuration.
- It doesn’t require developer expertise, reducing the cognitive load of having to learn something new. “So you don’t have to be an expert, you don’t even have to know what size database you should have. You can override our assumptions, but we’re going to provide all of that for you. So really, you don’t have to be an expert at all, you don’t have to know Terraform … you don’t have to know Kubernetes, although we’ll configure your Kubernetes cluster for you. We’re going to do all of those things on your behalf,” Kathman said.
- You can see value on Day 1. It doesn’t take weeks or months to learn and set up.
There are a number of competitors, like Terraform and Pulumi. Hathman also pointed to Harness, though he doesn’t consider it a direct competitor because it’s more enterprise-focused.
“At the same time, we’re doing the work that a Harness would do, but allowing an organization to deploy on Day 1, and so by automating, taking that opinionated approach and generating the IaC from the diagram, all of those things are just expediting the process significantly,” he said.
This article previously inaccurately said it was in public beta. No, it’s gone GA.
