Flume vs Timber

Flume was created to efficiently move log data to Apache Hadoop's HDFS.

Licensed under Apache 2.0.

The Timber interface stands out in the space:

1. It's beautiful, easy, and modern.
2. It's fast. It uses advanced front-end technologies to maximize performance and usability (react, redux, etc).

Timber offers a really great simple search, with a support for advanced features like: term negation, condition grouping (parenthesis), regex, field searches and conditions, etc.

Timber allows you to SQL query your logs just like you would a traditional database. The JOIN support is very powerful.

The retention Timber offers is exceptionally longer. The default is generally 2 - 4 weeks, Timber offers 6 months.

Because Timber fully embraces structured data, graphing is simple, real time, and fast. They provide a number of aggregates you can graph on.

Timber provides native libraries that automatically attach metadata to your logs. They've designed a schema that defines events and context, which normalizes the data and makes querying, alerting, and graphing simple and reliable.

Timber can be installed in < 1 min and usually with a single command. It asks your for your application details and provides 1 set of simple instructions.

Besides the free plan, Timber doesn't impose limits on the number of users, alerts, etc. It's entirely based on the amount of data allowed for the plan. It's refreshingly straightforward.

The alerts are real-time and the approach is thoughtful. Instead of blasting you with alerts every time it's triggered, they change the state of the alert once, notify you once, and then notify you again when the alert is no longer an issue.

Timber's docs are detailed and thoughtful, both for the service as well as the libraries they offer.

Beyond the space your plan is allowed to use, there is no rate limiting.

Logs show up < 3 seconds of when they are generated