Speakers
Description
Modern large distributed computing systems produce large amounts of monitoring data. In order for these systems to operate smoothly, under-performing or failing components have to be identified quickly, and preferably automatically, enabling the system managers to react accordingly.
In this contribution, we analyze job and data transfer data collected in the running of the LHC computing infrastructure. The monitoring data is harvested from the Elasticsearch database and converted to formats suitable for further processing. Based on various machine and deep learning techniques (clustering, supervised and unsupervised learning), we develop automatic tools for continuous monitoring of the health of the underlying systems. Our initial implementation is based on publicly available deep learning tools like the PyTorch or the TensorFlow packages, running on state of the art GPU systems.
| Consider for long presentation | No |
|---|
