Learn more about the presentations that make AutomaCon awesome.
Learn more about the presentations that make AutomaCon awesome.
It’s 2016, and "measure absolutely everything" is the new normal. Servers are cheap, and data is abundant. Time series data drives engineering and business decisions, and systems enabling such high-scale collection must scale along with the rest of the organization.
Over the years, Yelp accumulated many ways of generating metrics, monitoring, doing analytics, and alerting. In the past year, a dedicated metrics team formed to integrate and maintain all the tools as a platform for the wider engineering organization - providing monitoring as a service. To achieve this, prinicples from the Infrastructure as Code movement become essential: ephemeral servers managed within version control, distributed configuration management via CI/CD pipelines, and aggressive containerization of federated resources via PaaS techniques.
Join us to hear about auto-scaling stateful aggregation systems that don’t want to auto-scale. Listen in for tools that work across vendors, libraries supporting in-house innovation, and a strong push toward self-service metrics for all our internal users. Check out Yelp’s approaches to redundant alerting, cheap yet resilient storage, and expressive client libraries that actually don’t suck.
"Who monitors the monitors?" We do.