25, 2013, Quora released new versions of the site 46 times. This was a normal day for us.” - Quora / April 2013 “Deployment every 11.6 seconds, 1,079 max in one hour. 10,000 mean number of hosts per deployment, with 30,000 maximum” - Amazon.com / November 2012 “On the Google Consumer Surveys team, 8 minutes after you commit code it's live in production.” - Google / June 2014 “10+ deploys per day.” - John Allspaw / June 2009
deploy a change that involved just one single line of code? Do you do this on a repeatable, reliable basis? Continuous Delivery slideshare.net/jezhumble/adopting-continuous-delivery continuousdelivery.com
vintage items. Etsy.com employs 500+ people and has grown to a community of more than 25 million members and 1,000,000+ online shops who generated 1.35$ billion of sales in 2013. extfiles.etsy.com/progress-report/2013-Etsy-Progress-Report.pdf Deploying etsy.com 40+ times every day by a team of 175+ committers supporting 60+ million unique monthly visitors. youtube.com/watch?v=4pVOXLvoUyE infoq.com/news/2014/03/etsy-deploy-50-times-a-day slideshare.net/beamrider9/continuous-deployment-at-etsy-a-tale-of-two-approaches etsy.com
active retail partners worldwide. Between 100-150 services are accessed to build a page. fool.com/investing/general/2014/05/24/how-many-customers-does-amazon-have. aspx Deployment 1,079 max deploys per hour affecting 30,000 hosts simultaneous receiving a deployment. youtube.com/watch?v=dxk8b9rSKOo amazon.com
monthly releases, 2 week qa cycle, late night incidents, 12 month lead time → After: bi-weekly release, <1 day qa, single-button rollback, 1 week lead time
serving 94 of Fortune 100 companies. Third largest company in software revenue in the world. en.wikipedia.org/wiki/HP_Software_Division Deployment 1000+ builds per day, 825,000+ change requests in a year, 100,000+ tests run daily. h30614.www3.hp.com/collateral/Barcelona2014/presentations/B5756.pdf hp.com
Before: 30 days per feature delivery, 300min prod build, 1440min full test → After: 7 days per feature delivery, 10min prod build, 360min full test Ticketmaster - youtube.com/watch?v=01gfNOwhnro Before: Bi-monthyl/Quarterly releases → After: 109 deployments in September Target.com - youtube.com/watch?v=exrjV9V9vhY 30+ APIs, 80 deployments per week, <10 incidents per month
restore service change fail rate How long is the delay between a request for a change, and a production system operating with that change implemented? How long does it take for an abnormal behavior in the system to be restored to the normal standard agreed way of operation? How many changes and features are being released to production in a fixed period of time? How often the system fails or service is disrupted?
can bind a single interface, if the bind option is not # specified all the interfaces will listen for incoming connections. bind <%= @bind %> ... template "/etc/redis/redis.conf" do ... variables({ bind: node[:redis][:bind] }) end redis cookbook ... chef
version 11.4.0 Compiling Cookbooks... Converging 3 resources Recipe: redis::default * package[redis-server] action upgrade (up to date) * service[redis-server] action nothing (up to date) * template[/etc/redis/redis.conf] action create - update template[/etc/redis/redis.conf] from 81b4f1 to 8a6cec --- /etc/redis/redis.conf 2011-07-27 17:26:50.000000000 +0000 +++ /tmp/chef-rendered-template20130406-2537-f8vlv6 2013-04-06 ... @@ -27,7 +27,7 @@ # If you want you can bind a single interface, if the bind option is not # specified all the interfaces will listen for incoming connections. # -bind 127.0.0.1 +bind 0.0.0.0 # Specify the path for the unix socket that will be used to listen for # incoming connections. There is no default, so Redis will not listen * service[redis-server] action restart - restart service service[redis-server] Chef Client finished, 2 resources updated install package run service configuration + restart ... chef
Kyiv DevOps, September, 2015 - https://twitter.com/thoughtworks/status/658636712709480448 - 20th Century BIG IDEAS in Operations Management 1920s: Ford & Taylor - Moving production line and standardized work 1930s: Shewhart (& Deming) - Statistical Control of Quality 1960s: Ohno - Lean Production System 1980s: Goldratt & Kaplan - Measurement & Theory of Constraints 1990s: Hammer & Dell - Reengineering / Process Focus & “Direct Model”