DevOps culture: myths and reality
Clearing out the myths, then looking at what actually grew at scale
Slide contents
1. DevOps culture: myths and reality
Clearing out the myths, then looking at what actually grew at scale
2. The myths first, then what practice produced
What we'll cover
Myths: who, what, how
Reality: business and IT
Teams and maturity
Conclusions
3. 01. Myths
Who, what and how — three ways to miss
4. There is no such thing as a devops person
What people call one
Not a single person
Not an org unit next to dev and ops
Not the person who knows how to do everything
Not the old sysadmin, and not a dev who runs CI/CD
5. Devops gets swapped for its own form
What it gets passed off as
Not a set of the right fashionable technologies
Not another name for ITIL/ITSM
Not just a buzzword about development culture
6. Quick moves do not buy a culture
How to «do devops» from the outside
Consultants who preach about culture
Consultants who pitch fashionable technologies
Consultants who outsource CI/CD into a black box
Buying an external team wholesale
7. 02. Reality
How it started and what it turned into
8. Infrastructure grew into an internal product
One stack, two viewpoints
Physical infrastructure — what sits in the rack
Virtual infrastructure and IaaS — a resource on demand
PaaS and applications — what the business pays for
The business looks from above, IT from below
9. The development team owns its own production
The modern approach to building software
Separate services instead of monoliths
The team writes the code and runs it through CI/CD
The engineering team provides *aaS: PaaS, IaaS, DBaaS
And the tooling: logging, monitoring, service mesh, GitLab CI
10. Core and engineering merged into a platform team
Splitting into stream-aligned and platform teams
Development split into stream-aligned and core
Stream-aligned teams simply build business features
Core took on the shared tooling
Core and engineering merged to build the PaaS
11. 03. Teams and maturity
The Team Topologies reading
12. Each team type has its own customer
Four team types and three interaction modes
Stream-aligned team — the flow of business value
Platform team — an internal product
Enabling team — helps others adopt practices
Complicated-subsystem team — narrow expertise
13. A platform appears where scale and maturity meet
Engineering maturity against org size
Small scale: specialists and dense collaboration
More maturity: end-to-end ownership
More scale: specialists on top of a PaaS
Both at once: the focus shifts to reliability
14. 04. Conclusions
What to take with you
15. Devops is about joining development and operations
And about how that looks at different scales
Devops is an overloaded term
It is surrounded by a lot of myths
At its core it joins development and operations
At large scale: stream-aligned plus platform teams, with the platform teams building a PaaS
At small scale: teams are better off taking a cloud PaaS
16. What to read next
The links from the talk's closing slide
The «Team Topologies» book — with a three-part summary
An article on product and platform team collaboration
An article on the limits of reusing shared solutions
Stanislav Khalup's talk «DevOps eras at Tinkoff»
17. Thank you!
polomodov.tech
All slides and links — in the Telegram channel
Alexander Polomodov, Technical Director & Fellow, T-Technologies
@book_cube