William Brooks is not extinctNov 11, 2022
Vicki Boykis

Something I have been thinking about recently is the cycle of trade offs we made when moving to the cloud.

The premise was that we now don’t have to worry about hardware but as a result we now have to track infrastructure in our code. Often, this results in a pattern of writing code to AWS or GCP spec rather than writing the logic of our program. This pattern is brilliantly described in this post as “tending to the vendor’s stuff.”

https://rachelbythebay.com/w/2020/08/14/jobs/

We are a spectrum of jobs, not just one

Nov 11, 2022 at 12:53pmWeb
Show threadAlison Presmanes HillNov 11, 2022
@vicki “VendorOps” is so on point 
Show threadKristian GlassNov 11, 2022
@vicki Where does “on-prem mean no IaC” come from…? What about machine provisioning and orchestration etc?
Show threadVicki BoykisNov 11, 2022
@doismellburning that still happened but it wasn’t the developer’s responsibility: there is no way I could reasonably be expected to go into a data center and rack 😅. This also coincided with the shift to dev ops.
Show threadsanjayNov 11, 2022
@vicki very true. Really sucks that there isn't an Apache version of Terraform or something along those lines.
Show threadVicki BoykisNov 11, 2022
@sanjay i think the problem is not that there aren’t different versions of Terraform but that we need Terraform as a key part of our development cycle
Show threadsanjayNov 11, 2022
@vicki ah, I see what you meant. But would that imply if we need a non-vendor version of managing infrastructure, we are back to managing our own servers again?
Show threadVicki BoykisNov 11, 2022
@sanjay I think yes but that infrastructure is already mostly spun up and managed by SREs rather than devs. I wrote a bit about this: https://vickiboykis.com/2018/01/28/working-with-aws/
Working with AWS

Clouds, Isaac Levitan TL;DR: AWS is an extremely flexible environment to work in, but the flexibility can lead to an overwhelming amount of choices that may initially overwhelm and make someone used to working on-prem less productive until they get the hang of it. More and more of my projects lately have been either doing data science on AWS, or moving data into AWS for data science, and I wanted to jot down some brief thoughts on coming from an on-prem background about what to expect from working in the cloud.

★ Vicki Boykis ★
Show threadsanjayNov 12, 2022
@vicki this is very accurate. But I think could it also have to do with the fact that we are still early in a cloud based development environment? I'm not sure, just thinking out loud here. I have read about a project called Ray (https://github.com/ray-project/ray) that may be possibly what you are talking about but as far as I know it is limited to Python for now.
GitHub - ray-project/ray: Ray is a unified framework for scaling AI and Python applications. Ray consists of a core distributed runtime and a set of AI Libraries for accelerating ML workloads.

Ray is a unified framework for scaling AI and Python applications. Ray consists of a core distributed runtime and a set of AI Libraries for accelerating ML workloads. - GitHub - ray-project/ray: Ra...

GitHub