Let’s do things differently. To start with, let us view logs and traces as no different from any other data. The data an application indirectly generates when in use (the logs and traces) is no different from the data an application directly works with (input and output). So let’s keep them all together in a scalable cloud storage repository.

Once it is there, it is just like any other big data. We need to analyze and apply intelligent monitoring to detect situations of interest. So we need to apply trained ML models to a stream of such data for immediate alerting when the traces indicate an unwanted behavior occurring or brewing. This talk will show how to harness existing technologies to do just that.