Hi,
My understanding is that with very large Wireless LAN infrasturctures, it's better using the hash mechanism. When using the pooling mechanism, this implies building a lookup table to associate each client MAC Address to the VLAN where it is assigned. However, this also implies looking up in this table for every packet... This is not efficient.
The hash mechanism is much more efficient. However, you may end up with Wireless LAN clients not evenly distributed between your VLANs as this mechanism depends on the hash algorithm and the client MAC Address. Consequently, you may want to closely whatch IP utilisation for each VLAN and create more VLANs/subnets when required.
Best regards,