Role reuse: promising but hard - Martin Maisey.
- Why we love Ansible
- Why Galaxy doesn’t always work for us at the moment
- What makes a good role?
- How could we/Ansible Inc make Galaxy and Ansible’s metadata format better?
Presented at Ansible Meetup London 27/11/14 - see http://www.meetup.com/Ansible-London/events/218679344/
BJSS
Just over 21 years old –could say we’re – technically - a proper grown-up company now
Formed by co-founders who were annoyed at being managed by idiots, still privately owned, ethos remains
First gig was writing Solaris device drivers for Emulex HBAs, then got into trading systems, now do pretty much everything
Don’t ask what the name stands for, but we like to call ourselves a full service consulting company
Automating infrastructure well before the term DevOps existed
Used Puppet, mostly on large projects, for a number of years
Started adopting Ansible about a year ago, and it would probably now be a company standard if we had things like that
Redeveloped NHS Spine on Python / Riak / Tornado (El Reg passim.), went live recently without a fuss/hitting the front pages, saving the taxpayer a lot of money, providing a solid platform for new functionality
Also sent a major FX trading system from rebuild live, as a big bang, without anyone noticing – including reverse engineering undocumented terminal protocols and engineering in bug-for-bug compatibility
Both of the above built on DevOps principles from the start
Have been doing infra work for years, but formalised this four years back with an Infrastructure consulting group – now Infrastructure & Security
For a very large precision manufacturing firm, scope and managed a multi-year enterprise datacentre migration programme – including core network build-out, shared infrastructure, P2V, legacy application migration and updates to supported platforms, lift+shift of legacy boxes – for all mission-critical enterprise services
Before OpenStack resembled anything you’d want to touch with a bargepole