ConfPad

Today, no one is surprised by highly loaded systems. Moreover, few people in our industry can be surprised by a separate person who is engaged in stress testing. Most of them think if they can automate everything, then the tests would find bugs automatically. However, few people know that the most significant part of the time is spent on analyzing the results (logs and graphs). The speech is going to show how to reduce the time to find bottlenecks in the analysis of various logs. How can you apply the simplest models of machine learning to find bottlenecks? Moreover, how to get requirements using historical data.