25. 25
Traditional apps will not
• Drive your organization forward
• Allow you to innovate quickly
• Enable your staff to develop new skills with modern
tools
46. 46
Up to 60% of failures are caused
by human error,
not hardware failure
47. 47
Example - Knight Capital Group
https://infocus.emc.com/wp-content/uploads/2012/08/KCG.jpg
48. 48
• Manual deployment of new trading software
• 7 of 8 servers correctly updated
• Old function still alive on the 8th server led to…
Knight Capital Group - What happened?
78. 89
• Deploy, monitor and connect together all the pieces
needed to run services for the organization
• Processes and tools for faster end-to-end delivery of
quality services
• Automation comes built-in
Why define infrastructure as code?
79. 90
• Desired state specified in text files
• Autonomic (self-corrects to desired state)
• State should be known through monitoring
• Remove snowflake servers
How do we do this?
80. 91
• Easy to read and edit
• Shareable
• Can use standard version control like Git or SVN
• Becomes executable documentation
Why store them in text files?
81. 92
node 'www2' {
class { 'apache': } # use apache module
apache::vhost { 'awesomewebsite.com': # define vhost
port => '80',
docroot => '/var/www/html'
}
}
Simple example
84. 95
• Deploying, provisioning and scaling automatically is
virtually impossible if every server is unique
• Adds friction between the requestor and the deployer
• Mistakes happen
• We're all human
Snowflake servers
86. 97
Up to 60% of failures are caused
by human error,
not hardware failure
87. 98
Example - Knight Capital Group
https://infocus.emc.com/wp-content/uploads/2012/08/KCG.jpg
88. 99
• Manual deployment of new trading software
• 7 of 8 servers correctly updated
• Old function still alive on the 8th server led to…
Knight Capital Group - What happened?
92. 103
• Self documenting infrastructure
• You now have source code for how anything in your
datacenter is setup
• Executable documentation
Recap of the benefits
103. 114
• And by smartly we mean cooperatively between
Devs and Ops
• Will help operators understand what's running and
happening in the infrastructure
• Will give developers insight into the underlying
infrastructure and the way it behaves under load
Use these tools smartly
104. 115
• Start talking about it
• Choose low-hanging fruit
• Learn, make wrong decisions, learn again
• Involve the broader team
– Give books, presentations, blog posts etc.
So how do we get started?