days
-2
-9
hours
-2
0
minutes
-2
-5
seconds
-4
-7
No tricks, no fire, just an open source logging backend

Marvel at Grafana Loki: The Prometheus of open source log backends

Jane Elizabeth
Loki
Shutterstock / Asmus Koefoed

Marvel at Grafana’s latest open source offering for logs: Loki. No tricks here; this horizontally-scalable log aggregation system was inspired by Prometheus. Instead, Loki provides a cost-effective alternative for your logging backend system.

Looking for a new logging backend system? Loki is a highly-available, multi-tenant log aggregation system by Grafana. Inspired by Prometheus, this logging backend system is optimized for Kubernetes users, is 100% open source, and has never turned into a snake and then stabbed anyone as far as we know.

Loki provides developers with an efficient, cost-effective approach to log aggregation streams. It’s optimized for Grafana, the open platform for analytics and monitoring, as well as Kubernetes and Prometheus.

    Serverless Architecture Whitepaper

    LIVING IN A POST-CONTAINER WORLD
    Free: Brand new Serverless Architecture Whitepaper

    Stay on top of the latest trends with the help of our Serverless Architecture Whitepaper. Get deep insights on serverless platforms, cloud-native architecture, cloud services, the Kubernetes ecosystem (Knative, Istio etc.) and much more!

What’s up with Loki?

Loki is awesome and doing great. (Some of us refuse to acknowledge certain parts of the latest Avengers film, thank you very much.) It’s designed for efficiency and considerably different from other logging solutions, with native support in Grafana.

How? It doesn’t do full text indexing on logs. Instead, Loki stores compressed, unstructured logs and only indexes the metadata. This is because right now, logging is expensive to scale.

Instead of telling developers to “log less”, Loki allows developers to use all the logs they want, with the tradeoff that it won’t parse their log data. It’s meant to complement existing solutions like ElasticSearch or Splunk for the full text indexing. Since it only stores the metadata, Loki itself is cheaper to run and simpler to operate.

Loki uses the same labels for its indexes and groups log streams as in Prometheus, making it possible to streamline and seamlessly switch between the two.

Additionally, it automatically scrapes and indexes metadata like Pod labels, making it a cinch for storing Kubernetes Pod logs.

Loki consists of 3 components:

  • loki is the main server, responsible for storing logs and processing queries.
  • promtail is the agent, responsible for gathering logs and sending them to loki.
  • Grafana is for the UI.

Fair warning: Loki is an early alpha software. It’s not meant for production use (yet). Grafana is publishing it early in order to get valuable community feedback while everything is still fresh and malleable. I’m not saying that it’s unstable and will bring about the end of the world, but you probably shouldn’t trust Loki too much right now, much like his mythical namesake.

SEE ALSO: Logging in the age of microservices and the cloud

Getting Loki

Loki may be one of the slipperiest gods in the Norse pantheon, but trying out Grafana Loki is remarkably simple. Right now, there are five different ways developers can test out Grafana Loki. From easiest to hardest:

More information about Loki can be found on GitHub.

Author
Jane Elizabeth
Jane Elizabeth is an assistant editor for JAXenter.com.

Leave a Reply

Your email address will not be published.