OpenStack Summit May 2015 Vancouver

Openstack services create a big jungle of logs and metrics. These can provide a cloud operator tremendous insights and can help answer important questions, such as:

(1) What are the most interesting and actionable logs out of the many terabytes I see?

(2) Are there metrics whose changes correlate with certain logs?

(3) Are there certain errors that cascade from one service to another? How do I get to the root cause of issues?

(4) What are the typical parameters associated with common event pairs (average time between occurence, association rules, preceding and successive events)?

(5) Are there logs that are early warning signals for potential operational issues?

We shall also see which logs are indicative of anomalies and how visualizations of the anomalies can improve operational efficiencies