Assessing DevOps & Data Science and vice versa topicality

Question

First, to review the definitions from Wikipedia.

DevOps - DevOps is a software engineering culture and practice that aims at unifying software development (Dev) and software operation (Ops). The main characteristic of the DevOps movement is to strongly advocate automation and monitoring at all steps of software construction [..]. DevOps aims at shorter development cycles, increased deployment frequency, and more dependable releases, in close alignment with business objectives.

Data Science - Data science is an interdisciplinary field of scientific methods, processes, algorithms and systems to extract knowledge or insights from data in various forms, either structured or unstructured [..].

Now let's consider two cases.

A. DevOps for Data Science: use DevOps knowledge base to deliver shorter cycles in data science experiments, more automation and scalability as well as reproducible experiments. Business case: "works on my machine" is a known problem also in data science teams and transition from first experiments to production is not always easy.

B. Data Science for DevOps: use Data Science knowledge base to define and implement the vague "Measure" in DevOps principles. Using Industry 4.0 approach, you could evaluate data from the whole toolchain. Of course companies collect and analyse time series data, but afaik this is much more focused on production, or limited to the scope of separate tools. Business case: holistic data driven insights - not replacing but augmenting human interaction - in large scales environments (think of dozens of scrum teams and hundreds, even thousands of [micro]services).

Now my questions (please try to judge in terms of Q&A community strategy):

• Are A, B both more or less offtopic here or only one of them?
• Would you recommend proposing a separate community for either A,B or both?

Answer 1

I, personally (understand not as a devops moderator), see no reason why A or B would be off topic here.

We are using part of data science to create alerts on behavior change from application metrics (one of the most public attempt I know of is Skyline, the anomaly detection part of Kale by etsy.

Data Science formulas like standard deviation, median, meadian average on rolling period, are already in various tools like graphite/grafana or Elasticsearch. Elasticsearch 6 even include a "machine learning" engine, closely related to datascience.

So when the two topics interact, deploying a kafka cluster or using DS tools to analyse metrics, I see no reason to ban them from devops.se.

That said, there's probably a lot of questions about data science applied to a devops metric which are better suited to CrossValidated SE site as their relation to devops is not meaningful for the question when it's only about the statistical tools themselves as well as a bunch of cluster deployment things are as well suited to serverfault than here.

The fact something is on topic elsewhere on the SE network is not relevant to our own topicality, the main question remaining is: Are we ok and apt to answer deep statistical questions ?.