DevOps, SRE, DX: Getting clear on Terminology and Teams

Summary

Deprecate DevOps as a term
Service Reliabiliity Engineering (SRE) is preferred term, esp for the crew/team (which is led by Irio at present)
SRE covers 2 areas:
- DX (Developer Experience): creating and improving the systems and processes for efficiently creating and running reliable solutions
- Toil: day to day reactive support and maintenance
Support is a project. Product Owner is Irakli at present.
"You build it, you run it".

"X project has been launched and is entering hosting / support. Going forward SRE will own it."

"We need to deploy a new app. I will do this myself and only contact SRE if I run into problems"

"How can we improve DX in the area of continuous deployment."

Clarify and precision: DevOps is a bit imprecise and has stopped being developer ops and become a mix of various things. SRE is both more precise.

DX is also a useful new term.

Service Reliability Engineering (SRE) will be our term for the activity and crew that:

Ensure our systems run reliably, especially our hosted solutions for clients.
The Devloper Experience (DX) for our developers building and maintaining these solutions is great.

We will deprecate the term DevOps other than as a label for a type of task (deployment, maintenance etc) done by developers.

Deploying and maintaining applications is a responsibility of developers not a separate team: you build it, you run it.

At the same time, we need a "crew" who are:

High level experts (so escalation point for issues)
Design the DX – the systems and processses – that enable our developers to "run it" (and, to an extent, "build it")
Responsible for the running of the SaaS-like solutions (or solutions that have entered pure hosting mode)

SRE work includes 2 areas: DX and “Toil”

DX: Developer Experience = the experience of developers in carrying out their work, specifically in creating and managing DMSes and data driven applications. Can think of it as "Product" work for our internal systems/processes around development, deployment and management of services.
Toil: day to day support. Contrasts with DX "Product" work as it is interrupt driven and responsive.
- 9-5 or 24/7 cover
- Immediate response

Aim from Google book is that SRE team members spend less than 50% time on toil.

DevOps is something people (developers) do. It is not an area, team or role within the org. Literally it is Developers doing Operations.

SRE = Service Reliability Engineering. This is term we will adopt going forward for a specific “team” and area of responsibility.

We also have the concept of "support" (see http://playbook.datopian.com/support/). Support could usefully be divided into:

SRE: a system is down, not working. This may evolve into technical support if the issue is traced to a bug in the underlying solution.
Technical support: there is a bug in the solution (CKAN) that needs debugging and fixing.

Create [email protected] and [email protected] email addresses (former is catchall anyone can write to, latter is for team)
Rename devops project to sre
Send email to all-team
Talk through on all hands (?)

None