Report Open Access

Machine Learning applications on OpenStack log data analysis

Ravi Charan Nudurupati

A massive amount of data is generated by the Openstack cloud services in the format of service logs. Besides timestamps and log level fields, these logs contain additional information useful for pattern analysis. Unfortunately, this information is generally exposed in semi-structured text format, not allowing direct analysis without additional munging of the data. Traditional approaches to extract information from those fields are rule-based, mainly applying regular expressions upon knowledge of the text structure. These approaches require a pre-knowledge of all text patterns and are not scalable with the growth of the services. This report proposes a solution that is a mixture of the MinHash Locality Sensitive Hashing and the DB scan algorithm for data clustering. 

Files (1.2 MB)
Name Size
Report_Ravi_Charan Nudurupati.pdf
1.2 MB Download
All versions This version
Views 331331
Downloads 269269
Data volume 328.8 MB328.8 MB
Unique views 308308
Unique downloads 247247


Cite as