State of Monitoring today
● Products are increasingly being offered as a service to end customers. As a result the
deployment topology of the application becomes much more complicated. Tenancy and
isolation requirements are hard to relatively harder to meet. With the usage of
multi-cloud and multi-cluster deployments monitoring these applications and measuring
what matters becomes much more harder.
● Metric data needs to be isolated as well as preserved for a much much longer time to
analyze and identify patterns to reduce alert fatigue and get better with monitoring what
really matters.
● Companies increasingly find the need to adopt monitoring solutions such as Thanos,
Cortex, Victoria metrics in order to monitor large scale deployments which provide long
term storage, and are highly available in nature.
Krius
● Krius is a cli tool to ease the installation and configuration of Prometheus and Thanos across
multiple clusters without worrying about wiring them together using a single source of truth.
● Allows the end user to provide a complicated deployment topology using a declarative
specification file and orchestrates the bootstrapping of those components.
Krius UI
Built on top of Krius CLI tool to create and design the monitoring deployment topology across multiple clusters
Demo
● Showcase Krius UI
● Generate Krius Spec file - a Single declarative state of Monitoring State across multiple clusters
● Apply Spec file
● As a result, get an aggregated view of the deployment through Thanos Querier for all of your
storages like:
○ Prometheus (sidecars)
○ Metrics received from Prometheus remote write streams (Receiver)
○ Object Storage
○ Global alerting/recording rules evaluations (see Ruler)
Contributions welcome
● Read more about Krius -
https://www.infracloud.io/blogs/krius-accelerating-monitoring-adoption/
- https://github.com/infracloudio/krius