Intro to the Jut Dataflow Platform

December 11, 2014

Michael Demmer, VP Enginnering at Jut unveils their dataflow platform which has been under development for the past year and a half.

In this video Mike gives some demos of the system and shares how they think about modeling timeseries data along with unstructured events in a complex system.


  • Jut wants to tackle the problems of managing complex systems
  • Vs existing products which focus on one small slice of the pie
  • Data is diverse
    • We have regular metric data
    • …and also unstructured events like server logs and site actions
    • Historical and current events
  • Want to take all types of data, analyze it and automate alerts

Small Demos

  • Graphing 90th percentile server response times across various services
  • Single server response times overlayed with error events
  • Modularity of the Juttle language
  • Data joins
  • Sharing and loading Juttle gists within the Jut Playground

Data Integration

  • How to integrate in a SaaS/on-premise hybrid
  • Logs, metrics, events generated behind a firewall can be processed back there in a Jut “engine”
  • The UI and auth lives on, and it sends commands through your firewall via a control channel

Applications in Practice

  • Getting meta: Using Jut to monitor Jut usage itself
  • Example analyzing actual SSH break-in attempts on
  • Using the same program to view and mash up events gathered in different ways
  • Memory profiling of application
  • Q&A