Hey everyone! I'm a community engineer at Pulumi, and we're focusing on enhancing our Internal Developer Platform (IDP) features. I've been exploring various tools out there, but it's tricky to figure out which features are really beneficial beyond just flashy demos. From your experiences, what features do you think are essential for a successful IDP? I'm especially interested in how you manage the messy day-to-day aspects of infrastructure lifecycle management. Plus, if you have insights on using simple one-click provisioning setups, that'd be great too!
7 Answers
In my experience, the key to a solid IDP isn't just about the tech stack. It's more about establishing shared workflows and expectations among the team. Once everyone’s on the same page, the right solutions tend to emerge naturally.
What works for us is keeping it super simple. Our internal developer platform is just a CLI command: `./run-cli.sh deploy all/staging/demo/test`. This one command sets up everything from AWS to Proxmox, all linked with Ansible and GitHub Actions. I can kick it off, go grab lunch, and return to a fully deployed environment. The simpler it is, the more reliable it becomes—no more babysitting deployments or diving into complex UIs!
Haha, can you imagine doing this for SAP workloads? That would be a challenge!
Separating infrastructure from software lifecycles really streamlines things. It goes against the grain of what cloud providers often push, but it ultimately empowers developers to get the resources they need without overspending on infrastructure as code. It changes the game significantly!
We’ve just started using Backstage, thanks to management’s decision. It’s pretty neat for seeing all our services in one place, but honestly, developing and deploying through it feels a bit clunky. For most of our needs, GitHub templates get us about 90% there without the hassle.
Templates can't quite handle deployment or connecting services to external systems, you know? That’s where things start to get tricky!
I wonder why management pushed for Backstage in the first place. Did they really understand the platform needs? Are you guys using the scaffolder feature?
We've got a CLI too! It’s as straightforward as `ourCli start $foo`, which builds and starts the service. It even handles bootstrapping kind clusters and sets up all our necessary developer resources. It’s super handy and works for validating configs too!
The typical features everyone talks about for an IDP often lead to empty promises. For us, introducing scorecards and checklists for best practices really engaged our developers. Teams earn points for following guidelines, and now they actively look for ways to improve and request new features for the IDP once they saw tangible benefits.
That’s an interesting approach! I would have never guessed that would make such an impact.
An IDP truly excels with deep custom integrations to cater to unique processes. However, achieving this with a single click is quite the feat—it's challenging to build but can drive significant improvements.
That sounds intriguing! Could you share any specific integration examples you have in mind?
I love this approach! What led you to choose Swarm? And can you share what’s in your `run-cli.sh`? Is it just bash scripting or do you wrap other tools there? Also, how do you handle lifecycle changes, like migrating instance types? Is it built into the CLI or more manual?