12-01-2012, 04:31 PM
Dynamic load management of virtual machines
[attachment=16037]
1 Introduction
Existing data centers are characterized by high operating costs, inefficiencies,
and by myriads of distributed and heterogeneous servers that add complexity in
terms of security and management. In order to improve data center efficiency,
most enterprises are going to consolidate existing systems through virtualization
solutions up to cloud centers. Logically pooling all system resources and centralizing
resource management allow to increase overall utilization and lowering
management costs. There are various approaches to virtualization (hardware
virtualization up to micro-partitioning, operating system virtualization, software
virtualization), but consolidation and virtualization by themselves do little
to improve application performance. The question is whether huge increases in
terms of system utilization correspond to an actual better efficiency or they are
due to applications running poorly in those virtual environments.
Sara Casolari et al.
performance analysis and runtime management in these contexts are becoming
extremely complex, because they are a function not only of guest applications,
but also of their interactions with other guest machines as they contend for processing
and I/O resources of their host machine. We should consider that these
modern cloud infrastructures must accommodate varying demands for different
types of processing within certain time constraints, hence dynamic management
of virtualized application environments is becoming very important
Related work
There are several proposals for live migration of virtual machines in clusters of
servers, and the most recent techniques aim to reduce downtime during migration.
For example, the solution in Clark et al. [6] is able to transfer an entire
machine with a downtime of few hundreds of milliseconds. Travostino et al. [7]
migrate virtual machines on a WAN area with just 1-2 seconds of application
downtime through lightpath [8]
3 Management algorithms for load migration
A typical cloud architecture consists of a huge set of physical machines (host),
each of them equipped with some virtualization mechanisms, from hardware virtualization
up to micro-partitioning, operating system virtualization, software
virtualization. These mechanisms allow each machine to host a concurrent execution
of several virtual machines (guest) each with its own operating system
and applications.
[attachment=16037]
1 Introduction
Existing data centers are characterized by high operating costs, inefficiencies,
and by myriads of distributed and heterogeneous servers that add complexity in
terms of security and management. In order to improve data center efficiency,
most enterprises are going to consolidate existing systems through virtualization
solutions up to cloud centers. Logically pooling all system resources and centralizing
resource management allow to increase overall utilization and lowering
management costs. There are various approaches to virtualization (hardware
virtualization up to micro-partitioning, operating system virtualization, software
virtualization), but consolidation and virtualization by themselves do little
to improve application performance. The question is whether huge increases in
terms of system utilization correspond to an actual better efficiency or they are
due to applications running poorly in those virtual environments.
Sara Casolari et al.
performance analysis and runtime management in these contexts are becoming
extremely complex, because they are a function not only of guest applications,
but also of their interactions with other guest machines as they contend for processing
and I/O resources of their host machine. We should consider that these
modern cloud infrastructures must accommodate varying demands for different
types of processing within certain time constraints, hence dynamic management
of virtualized application environments is becoming very important
Related work
There are several proposals for live migration of virtual machines in clusters of
servers, and the most recent techniques aim to reduce downtime during migration.
For example, the solution in Clark et al. [6] is able to transfer an entire
machine with a downtime of few hundreds of milliseconds. Travostino et al. [7]
migrate virtual machines on a WAN area with just 1-2 seconds of application
downtime through lightpath [8]
3 Management algorithms for load migration
A typical cloud architecture consists of a huge set of physical machines (host),
each of them equipped with some virtualization mechanisms, from hardware virtualization
up to micro-partitioning, operating system virtualization, software
virtualization. These mechanisms allow each machine to host a concurrent execution
of several virtual machines (guest) each with its own operating system
and applications.