[one-users] (anti-) affinity groups support for scheduler

[Stefan Kooman]

Quoting Carlos Martín Sánchez (cmartin at opennebula.org):
> Hi Stefan,
> 
> How do you plan to implement it? I was thinking that instead of making the
> scheduler go through all the VMs running on a Host, the core could have
> support for this. It could add/remove the value of VM's AFFINITY_GROUP to
> the host on deployment/shutdown.
> 
> This way the host will have AFFINITY_GROUP="nameA, nameB, nameC"
> automatically populated, making the scheduling faster.

Support in core would be even better. At first I was thinking about a
similar approach: make something that runs on the hypervisors and create
"AFFINITY_GROUPS" based on some info. But it's hard to make it
"dynamically" as there is no coupling between the scheduler and the
hosts a priori. Having support for this in core would be awesome, and
would speed op scheduling. Proably scales better too: instead of
checking for a (number of) vm's on all hosts it's reduced to checking
for an "attribute" at all hosts. 

Gr. Stefan

-- 
| BIT BV  http://www.bit.nl/        Kamer van Koophandel 09090351
| GPG: 0xD14839C6                   +31 318 648 688 / info at bit.nl