4. Virtual Infrastructure Management
• Xen
• KVM
• VMware
Uniform and
homogeneous
view
configurable
resource
allocation
policies
• High Availability
• Server consolidation to minimize power usage
VIE
• Setting up networks dynamically for groups
• VM disk image deployment
• On-the-fly software environment creation
VM’s full life
cycle
changing
resource
needs
• peaks in which local resources are insufficient
• changing resources(addition or failure of
physical resources)
6. The Current of Cloud Ecosystem
Vmware
vSphere
Platform VM
Orchestrator
Ovirt
Current VI
management
tools
1
• Dynamic placement
2
• VM management on a pool of physical resources
3
• automatic load balancing
4
• server consolidation
5
• dynamic infrastructure resizing and partitioning
8. The OpenNebula Architecture
Image and storage
technologies
VM’s Life Circle
Support services
deployment
OpenNebula
Network fabric
Underlying hypervisors
for creating and
controlling VMs
Separate scheduler
component
Interfaces to other data
center management
tools
Support a hybrid cloud
model
11. The Haizea Lease Manager
Act as a VM scheduler for OpenNebula
Role
Be used on its own as a simulator to evaluate different scheduling strategies’performance
over time
Lease : hardware resources, software environments, the period
Better-offer
Placement
policies :
AR
immediate
Implement: VM
Introduce runtime overhead
solution
12. The Cloud Ecosystem
The Haizea project
Act as a drop-in replacement for OpenNebula’s default scheduler.
By working together, OpenNebula could offer resource leases as a fundamental provisioning
abstraction, and Haizea could operate with real hardware through OpenNebula.
Integrating OpenNebula and Haizea provides the only VI management solution offering
advance reservation of capacity. As Table shows, other VI managers use immediate provisioning
or best-effort provisioning. However, private clouds — specifically those with limited
resources in which not all requests are satisfiable immediately owing to lack of resources —
stand to benefit from more sophisticated VM placement strategies supporting
queues, priorities, and ARs.