Datacenters are increasingly viewed as just a large pool of resources, such as CPU cores, RAM and disk space.
Managing these resources on individual machines becomes difficult as more and more applications should be run on top of this pool of resources.
Effective algorithms need to be put in place to optimize the usage of the available resources, while dealing with contraints such as specific security, or locality/non-locality requirements for some scheduling requests. After all, its ususally undesirable to run, for instance, services of the same kind or services with different security levels, on the same physical host.

On the web