I plan on trunking my switches (3 hosts per cluster) and teaming 2 nic's across each switch to ensure full redundancy.
That will be fine.
Additionally I plan on using route based by IP hash.
This, however will only work if your physical switches support some sort of multi-chassis LAG or similar and present themselves to the hosts as one logical switch.
Otherwise you will get in trouble because the host has no idea it's talking to two different switches. Likewise, the physical switches too have no idea that there are actually two other ports participating in their LAG. Expect crazy MAC flapping and maybe an entire network meltdown.
If your switches do support MLAG or similar technique and you really want to go that way, then you should use an LACP-based LAG.
Instead of using IP-hash/LAG, I would use load based teaming on the dvS unless you have really specific requirements that make you think you absolutely need LAG.
You should read these articles on the pros and cons of using LAG or not:
http://wahlnetwork.com/2014/01/13/vsphere-need-lag-bandaids/
http://wahlnetwork.com/2014/02/05/revenge-lag-networks-paradise/
http://wahlnetwork.com/2013/03/05/stop-using-port-channels-to-vsphere-hosts/
--
http://alpacapowered.wordpress.com