Monitoring capabilities of any large-scale distributed environment
represent a fundamental precondition for successful operation of such
environment. Monitoring information is continuously used in numerous
ways, including but not limited to: accounting, audit tracking,
debugging, fault detection, job scheduling and performance analysis.
As cloud computing is growing in popularity it is becoming apparent
that existing approaches to monitoring are at times insufficient. The
goal of this talk is to present several basic requirements that
current solutions do not offer or only partially. These requirements
are analyzed in detail and advocate their necessity for modern (cloud)
monitoring system. In addition a possible approach and general
architecture of new generation event-based monitoring system is
presented. Resulting prototype implementation is then shortly
described and evaluated.