It failed miserably:
$ tf graph|dot -Tpng > graph.png
dot: graph is too large for cairo-renderer bitmaps. Scaling by 0.420618 to fit
Trying to upload the genrated image to Google Slides and the error message: The image is too large. Images must be smaller than 25 megapixels.
$ tf state list|cut -d. -f3|grep ^google_|sort|uniq|wc -l
42 <-- different types of Google Cloud resources we use in our Terraform configs.
I'm very skeptical this config can be managed using a WYSIWYG tool.
People at hashicorp are aware of it but it doesn't seem to be a priority getting the output to look nicer right now
All for better visualizations of terraform resources tho
This feels like the G-Programming language from LabView which allows you to graphically program your physics experiments. But once you drink the programming language coolaid you don’t want to do graphics programming any more.
So I guess I am already biased towards reading text. But for me this is still the best way to consume information compared to pictures or videos
Is this a thing you have experienced for yourself and others around you? My experience does not hold this statement as it is much harder for me to parse a complex image.
This file is large (15.7MB) and took 1.2 minutes to load
So your website looks broken for 1.2 minutes for me.
I doubt anyone else but a web-dev has the patience to wait for it.
It’s like all the visual programming languages that become unusable as soon as the program achieves a bit of complexity.
Indeed, managing complexity is one of the most difficult part of making a graphical tools useful, and some architectures cannot inherently be represented only by a simple 2d graphic.
On the flip side, IAC can sometime be complicated to reason about without deep-diving in the source code and having deep knowledge about Terraform and every cloud provider.
We believe that there is a middle ground, where Brainboard helps engineers design and communicate about their architecture efficiently, are able to manually edit the generated Terraform files, while avoiding the churn of constantly redrawing their documentation graphics.
Anyways, being able to visualize infrastructure this way is pretty great. Currently, the way it usually works is to draw the infra and then manually write the tf resources. This seems like it would simplify the process greatly since one could genereate the IAC directly from the diagrams. Pretty neat (if it works well).
Looks interesting. I'll try to break it with the complexity of a medium-sized project next year.
