Horacio,
Let me put it this way: if the only device connected to a switchport is the IP phone, then it is largely irrelevant (with respect to the IP phone operation) whether you configure the port with a data VLAN (switchport access vlan data-vlan) and a separate voice VLAN (switchport voice vlan voice-vlan), or whether you configure the port directly into the voice VLAN (switchport access vlan voice-vlan). In the second case, you would normally also use the switchport voice vlan dot1p or switchport voice vlan untagged command to let the IP phone know rapidly that the voice VLAN is equal to the data VLAN, just the tagging is either allowed for QoS purposes (dot1p) or is not allowed at all (untagged). Having a different data and voice VLAN is necessary if you expect both a PC and an IP phone to be connected to the same switchport (a PC connected to an IP phone, and the phone in turn connected to a switchport). In that case, you most probably want to keep the data communication separate from the voice communication, which is why you can define two separate VLANs on a single access port.
An advantage of using an explicit voice VLAN may be that the phone will be sending tagged frames and it will be able to use the CoS priority field in the tag to perform QoS marking. Even without the tags, though, the phone will be marking the IP packets with appropriate DSCP values, and Cisco Catalyst switches are capable of performing QoS operations based on IP DSCP field. Therefore, using the tags just for QoS purposes is not a persuasive reason. You'd use different VLANs simply because you want to have your network neat and tidy and not mix different kinds of traffic and its purpose in a single VLAN.
Best regards,
Peter
