Last active
August 17, 2022 23:46
-
-
Save vinyar/a88ffbd93ee6705bd97c85017054d0b9 to your computer and use it in GitHub Desktop.
DevOps presentation notes
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
DevOps | |
Wrong ways to think about DevOps: | |
There is no devops.exe - you can't 'install' DevOps | |
It's not a THING, it's a way of getting things done | |
You can't ask the team to 'go get some devops' | |
DevOps is much more than just Developers and Operations. The term has long ago transended dev and ops. Evangelists know this, but not everyone else | |
DevOps is a transformational process | |
Messaging: | |
DevOps messaging on the sales side is different to Practitionlers (Dev, Ops), Leadership, Champions | |
Change comes from groundswell and leadership backing | |
Need to have buyin from people doing the work, can't mandate "devOps" (collaboration) - have to get people doing the real work excited | |
Pushing DevOps down doesnt work - People doing the work will say 'here comes another bull shit mandate' | |
https://www.arresteddevops.com/devops-culture-change/ | |
On the flip side, people doing the work won't always be equiped to sell it to leadership | |
How do you Sell DevOps? | |
Talk about results: time to market, ci/cd, metrics | |
devops is bathwater in which all of these things are babies. | |
Tools can faciliate devops | |
DevOops - what exactly are you trying to solve? | |
ask more questions | |
drill into what "we want DevOps" means | |
DevOps is a bunch of lego pieces that you as a professional arrange to solve company specific problems - Alex | |
Adoption: | |
devops is an extension of the agile movement (per many industry leaders) - shot itterations, cycles, etc. | |
mindset that it all leads to devops - "once you see the way, you see the way in all things" | |
Dont meet people half-way. Meet them where they're at. | |
Where can you learn more about DevOps & SRE? | |
Events. | |
Meet people inside your company who are teaching devops inside your company | |
you can't sit behind your desk and learn devops. | |
Hands down - meetups and your local devops days will be the place where you learn the most. | |
Listen, share, do the 'hallway track' | |
Get to bigger conferences | |
Make friends. Meet real practitioners. You can get inside peoples heads, get feedback on your thoughts. | |
Blog posts are good, but it's just not the same. | |
Bridget - DevOps is not something you'll find on a balance sheet | |
Does Operations engineer do DevOps? Sure. | |
How about support engineers when they deal with tickets? Sure | |
------------------- | |
WTF IS SRE:: | |
SRE is the natural evolution of DevOps | |
DevOps - Bringing application developers and Operators together | |
SRE - moving the conversation from local optimizations to larger systems thinking | |
understanding and optimizing modern applications across the entire (eco)systems | |
from security practices, to observability, to compliance, to culture | |
Security - DevSecOps | |
Observability - core of SRE | |
Multi-cloud | |
Historically - In my experience - DevOps has been about breaking down the silos, or perhaps just poking a tiny hole,to fit a | |
straw in, and over time making that hole big enogh for a handshake. This aspect of devops is often about building a ground | |
swell, working with engineers. | |
A lot of this work takes space during your discressionary time, or initially as part of keeping the lights on, really it's all | |
one big shadow IT project working AROUND the process and incompetent implementation of process tooling. Over time, it's | |
nurturing that relationship into a hole big enough for a handshake, and eventually breaking down the wall completely. | |
SRE - is much more about shifting things left. Automation and chaining processes is almost taken for granted. | |
Shift learning left as well. | |
SRE and observability is about learning how your product will behave when it fails BEFORE it fails. | |
< Add cost of learning graph > from cost of detecting failures in test vs prod. | |
If you can get very good and very quick at detecting failures, you can change quicker. | |
This is where scrum comes in. If you can change cheaply, and safely, you can gain a lot of benefits and knowledge. | |
............... | |
The code looks like it works | |
how will you konw it doesnt? | |
how will you operate it in production? | |
............... | |
In my experience: | |
Legacy devops is more about pets and building bridges | |
Modern devops is more about ephemeral and immutable automation and building bridges | |
SRE is moving towards resilience and recovery at scale, and just as much about building bridges as devops is | |
In either case, infra or code, you need monitoring and metrics. | |
Always. | |
Without metrics and metrics, it's clueless leading the helpless. | |
Supporting infra that's ready to die (die=distributed, immutable, ephemeral) is a lot easier that feeding and caring for a pet | |
immutable - cant change the system once it's been deployed. No SSH, no kubectl exec, no changes on disk. | |
once a system is deployed, you can't touch it. | |
all of your changes are in source control | |
Ephemeral - short lived infra that serves the purpose and dies. Short lived secrets. No major planning needed. | |
your infra is cattle, but your passwords are pets? bad | |
< another idea > - who is responsible for these aspects? DevOps or SRE? Either. Whoever can build it, help architect it, and train others. | |
............... | |
Things every should know | |
security - everyone is competent enough and knows enough. | |
response - everyone knows where it's written down, and knows how to execute it. It's been used and tested. | |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
(camels nose - if you allow the camels nose into the tent, pretty soon you'll have the whole camel in the tent)