Immutable infrastructure has changed the way we think about system lifecycle: compute machines live for days instead of months or years, and applications live for hours or less. With the proliferation of CI/CD systems, and infrastructure as a service, the increased churn in production systems has hastened the immediate need for tools that prioritize experimentation - is your next development iteration really better than the last? In such a volatile world, traditional notions of compute “environments” and mutable approaches to experimentation can be found lacking. In large systems, emergent behaviors are near impossible to replicate in isolation, so experimenting in production systems is the only way to effectively measure hypothesis. This session covers different schemes for experimentation and the primitives required to make converged infrastructure work for real systems.
16. – Conway’s Law
“Any organization that designs a system will produce a design
whose structure is a copy of the organization's
communication structure”
17. – Hyrum’s Law
“With a sufficient number of users of an interface,
all observable behaviors of your interface will
be depended upon by somebody.”
49. Challenges.
• How you collect results affects how you read them.
• Twyman’s Law: “Any piece of data or evidence that looks interesting
or unusual is probably wrong!”.
• Disparate data collection in many organizations.
• Instrument everything with segment identifiers.
• Modify all applications to be experiment aware.
• Such systems breed organizational change. Embrace it.
50. Conclusions.
• Experimentation infrastructure makes your workloads mobile.
• Not only does this make experimentation viable and low-impact,
but it makes your operation vastly more flexible.
• Codify and automate an experimentation process.
• You’re already doing it manually.
• Tools are already available that form the building blocks.