Expanded tips for deploying to AWS?
rsignell-usgs opened this issue · comments
Do you guys have plans to expand this Deploying to AWS section?
A step-by-step recipe, Blog post or even a few more details would be awesome.
Or maybe easier: a short video walk-through?
We use the Holoviz-backed-by-Bokeh suite of tools extensively on the Pangeo project to interactively explore remote sensing, climate and forecast model output and other large data.
A common desired use case is deploying an dashboard for exploring these datasets to a larger community, often on AWS.
I think the entire community would find an expanded instructions very useful!
(but I realize you have many very useful things already on your to-do list...)
All of those things would be great @rsignell-usgs we mostly just need more help/resources.
Also, regarding that linked section specifically, I guess it's not entirely clear to me that Elastic Beanstalk would be the common choice for most users. I had reasons for choosing it several years ago but there are definitely more/different options now. But I don't really personally feel like I have the necessary expertise to evaluate them all or make recommendations.
@bryevdv, we ended up deploying our Bokeh (Panel) Apps on AWS using ECS w/Fargate, using CDK scripts with GitLab runners to build the container and deploy the infrastructure. We based our Dockerfile off of your Bokeh example Dockerfile, and used CDK typescript for deployment.
It looks like the approach described in this AWS blog post using Terraform to deploy ECS/Fargate would be a more flexible alternative!
I've had people suggest Terraform/Fargate as the current best practice but I have zero experience with either
@bryevdv , let's hope that if people have success using Terraform to deploy bokeh/panel apps they report back here! 😸