[one-users] Load balancing in OpenNebula

[Luis Rodero-Merino]

Hello,

We would appreciate if we could get some info regarding any load
balancing functionality that could be available in OpenNebula.

1) At application level, does OpenNebula provide some load balancer
(e.g. http load balancer) at to distribute load among users' VMs? Or
such component should also be provisioned by the users themselves?
2) At VMs/hypervisors level, does OpenNebula implement any policy to
decide how to deploy VMs on the available hosts? For example, a) it
could try to deploy as many VMs on as few hosts as possible to
decrease resources usage; b) it could try to 'spread' VMs among the
maximum number of hosts to increase the resiliency to hw failures.
Dioes OpenNebula use any of those policies (or any other in
particular)?
3) Again at VMs/hypervisors level, OpenNebula can use Haizea to
implement the concept of leases. But is there any mechanism to ensure
that no user gets (many) more resources than others (for example by
flooding the system with petitions for leases)? Also, it is planned to
add to Haizea the possibility "to let developers specify their own
resource selection policies", which we guess could be used also to
implement fair resource-sharing among cloud users. Is that
functionality already available?

Thanks for your help and patience,

-- 
----------------------------------------------
Dr. Luis Rodero-Merino
Graal/Avalon Group, INRIA-ENS Lyon, France
----------------------------------------------