I know that Cisco gear can do this, but can a Procurve be made to do it as well?

I have a scenario of around 16 access switches, one switch per building, connected to two aggregation/core switches all via a redundant set of 10 Gbps uplinks. Each switch has one link to one of the two cores. I was thinking of creating multiple VLANs per access switch (one per floor), then using InterVLAN routing on the switch for inter-floor communications, then aggregating traffic on the upper half of the floors on the 10GigE link going to Core1, and the lower half going on the 10GigE link to Core2, with said VLANs failing over to the other link in case a core goes down.

The core switches handle InterVLAN communications between buildings, and the two HyperV nodes connect directly via two 10GigE links to the core, one link to one core switch. Ideally I'd do load balancing as well.

I'm used to cisco gear, so I'm not really sure if this is feasible or even recommended. Any tips?

EDIT: To clarify – everything is procurve, both access and aggregation switches. From what I understand, if STP is enabled and you have two links to a root switch, one is set to forwarding and one is set to backup per VLAN, which can result in all VLAN traffic "congealing" to a single 10GigE link. I would like to configure differing root switches per VLAN so that the traffic across the network is semi-distributed across the two links, basically balancing by source subnet.

Additionally, each switch has a link to the Hyper-V cluster nodes (2 nodes). I believe LACP on 10GigE NICs for Windows Server 2008 R2 is still a bit problematic, so I plan on manually tuning things, unless of course LACP for Intel 10GigE cards on Procurve 5406'es are okay now?