Logging and NoOps with Christian Beedgen

“You write the code, but you don’t run it? That’s just preposterous.”

Software applications are constantly generating logs. These logs are necessary to understand how an application is functioning, and logs are key to debugging. As applications have gotten more complex, logging infrastructure has become complex as well. Storing and managing all of our log data is such a big task that several companies have been started to tackle this problem.

Today’s guest is Christian Beedgen, CTO at Sumo Logic. Sumo Logic is a cloud-based log management company. We discuss the elastic log-processing platform Sumo Logic has built to help software engineers with log management. It’s a great conversation about distributed systems, machine learning, and debugging applications.

Questions

• How has logging changed for applications in the cloud?
• What do you mean when you say “log data is big data”?
• What kinds of elasticity are important in your log architecture?
• What is the ingestion path for the logs?
• Are there any interested distributed systems problems at Sumo Logic?
• Why do you want to replicate three times?
• How do you use machine learning to improve log management?
• Can you describe what NoOps is, and how it reflects Sumo Logic’s culture?

Links

• Sumo Logic
• tail
• Elastic log processing (video)
• Multitenant architecture
• Inverted index
• Lucene
• JSON logs
• NoOps Debate Grows Heated
• Christian on Twitter
  

Pranay