Monitoring in a Virtualized Environment

Abstract

Monitoring solutions for virtualized
infrastructure (VI) should evolve to collect, analyze and
provide configuration recommendation based on a broader
range of operational metrics. A virtualized infrastructure
is a complex interaction of hardware (servers, network and
storage), hosting variety of multi-tier application with specific
service level requirements and governed by their security and
compliance policies. Most existing solutions of today monitor
and analyze only a subset of these interactions. The analysis and
recommendation obtained tend to optimize only particular aspects
of the infrastructure and can potentially introduce violations for
the others. A virtualized infrastructure is dynamic in nature,
providing immense opportunities to automate configuration
changes to virtual machines, networks and storage. It delivers
the capability to administer the whole of infrastructure as a large
resource pool shared by multiple workloads. Monitoring solutions
that look at only few aspects end up forcing administrators to
create silos within the infrastructure that are specially designed
to ensure that business service requirements are met for the
specific applications running there. A monitoring solution that
can collect and analyze multiple aspects for assisting in decision
making and process automation can deliver greater efficiency to
the virtualized infrastructure.
In this paper we argue the importance of having a
monitoring solution that provides a holistic view of the
virtualized infrastructure. We discuss the need for solutions to
be capable of monitoring and analyzing a broader set of metrics
such as health of infrastructure components; performance of
operating environment such as hypervisors, operating systems
and application running on them; capacity utilization indicators
for server, networks and storages; information available with
configuration and change management database containing
policies including security and compliance policies. We also take
a look at what these broader set of metrics are and who would
be interested in them. The paper further proposes a monitoring
framework for collecting and analyzing the above mentioned
aspects of a virtual infrastructure to develop a more complete
solution.