Loki compared to other log systems


Loki / Promtail / Grafana vs EFK

The EFK (Elasticsearch, Fluentd, Kibana) stack is used to ingest, visualize, and query for logs from various sources.

Data in Elasticsearch is stored on-disk as unstructured JSON objects. Both the keys for each object and the contents of each key are indexed. Data can then be queried using a JSON object to define a query (called the Query DSL) or through the Lucene query language.

In comparison, Loki in single-binary mode can store data on-disk, but in horizontally-scalable mode data is stored in a cloud storage system such as S3, GCS, or Cassandra. Logs are stored in plaintext form tagged with a set of label names and values, where only the label pairs are indexed. This tradeoff makes it cheaper to operate than a full index and allows developers to aggressively log from their applications. Logs in Loki are queried using LogQL. However, because of this design tradeoff, LogQL queries that filter based on content (i.e., text within the log lines) require loading all chunks within the search window that match the labels defined in the query.

Fluentd is usually used to collect and forward logs to Elasticsearch. Fluentd is called a data collector which can ingest logs from many sources, process it, and forward it to one or more targets.

In comparison, Promtail’s use case is specifically tailored to Loki. Its main mode of operation is to discover log files stored on disk and forward them associated with a set of labels to Loki. Promtail can do service discovery for Kubernetes pods running on the same node as Promtail, act as a container sidecar or a Docker logging driver, read logs from specified folders, and tail the systemd journal.

The way Loki represents logs by a set of label pairs is similar to how Prometheus represents metrics. When deployed in an environment alongside Prometheus, logs from Promtail usually have the same labels as your applications metrics thanks to using the same service discovery mechanisms. Having logs and metrics with the same levels enables users to seamlessly context switch between metrics and logs, helping with root cause analysis.

Kibana is used to visualize and search Elasticsearch data and is very powerful for doing analytics on that data. Kibana provides many visualization tools to do data analysis, such as location maps, machine learning for anomaly detection, and graphs to discover relationships in data. Alerts can be configured to notify users when an unexpected condition occurs.

In comparison, Grafana is tailored specifically towards time series data from sources like Prometheus and Loki. Dashboards can be set up to visualize metrics (log support coming soon) and an explore view can be used to make ad-hoc queries against your data. Like Kibana, Grafana supports alerting based on your metrics.