Dev and Ops and System Ops oh my

From CitconWiki
Jump to: navigation, search

Unedited notes


concrete steps to get people to work together


TIM Group

Already quite agile but there are always potentials for issues. Once an outage happened and they wanted to work together but didnt know what the first concrete step to get people to work well

investigation to work


Want to not have the outage in general

Lots of internal outages - if something painful then do it more often - disaster recovery

JTF - Jeffery - doing archeology on our systems

Tom 100 developers 2 ops guys who wrack servers and install the OS - devs had to do everything else No TDD / pairing or other agile stuff Devs knew that they would get shouted at all all hours of the night then they wrote better code - when people feel (share the pain) then they do better

Disaster recovery sessions - opportunities to learn


Tim group ops team managed the hosting centre and developers has no visibility of what was going on. - lack of viability whole staging environments

Run your tests against production environments

Change freezes


deploy to test environment and test and fix in that the test environment

Human aspect - are they incentivised to work together to work on the right priorities - 1st production works then work on something eles - having a fire warden - compartmental issues can cause people to devolve responsibility - need everyone to align priorities to that of the business

- hard to get people to listen - align things to their own experiences - like not getting woken up at 3 am in the morning


Resolve technical debt by getting devs to do devops/sysadmins stories


Cross functionality standups


No evidence of an operations team - there are build monitors but no system admin visualizations - organisation needs to make visualization of support and devops a priority


Devops is a continuum - Benjamin Mitchel - make operational metrics available to the development team - they can feel the impact of the changes they are making

non-functional assertions about your application

deployable applications - using

accepting testing definitions - chaos monkey - netfix - goes around breaking things in production - agreement up front


fire drill training people on how to do post-mortem of the applications - choosing something that will go down and making it everyones responsibilty to fix it.

Dont throw code over the wall - make everyone responsible


Continual learning - envolvement of new insight


Organisational debt


Designing a new component - using mongo db - then when it comes to deployment then you tell the Devops about Mongo db - tell them

Delivery architect - talk about the architecture as the development


Sysadmin - limited provisioning for non-production systems


Same Physical location - people and / or information

Why is there no ops team members in with the devs - what would they concretely do ? - - help the environments be balanced


Mental model


Co-ordination between devs and ops - make sure app deployment works all the way through the staging environments - deployment process should be the same for all environments



Devops work - security patches, OS upgrades, etc - enviroment changes - deployment improvements - pairing / Brown bags to share knowledge - gaining metrics and visualisations - visualising devops task between teams

Read the book Slack