|You need a LACP capable switch in order to play with link aggregation. Unless you're talking a high capacity database type server that incurs a lot of access, you won't need the combined bandwidth and with a single switch you have no redundancy as your single switch is still a single point of failure.|
Besides, as tonysathre pointed out, if your storage doesn't have similar capacity, you're wasting your time.
For example, lets say you have an L2, LACP capable switch that's 1000 Mbps. You decide to do a 4 port LAG (link aggregation group) between it and the server. You configure the 4 switch ports and the 4 interfaces on the server and hook everything up and you're now getting a combined throughput of up to 40000 Mbps from server to switch. If your storage (say a NAS) has only a single 1000 Mbps connection to the switch, you lose any/all bandwidth benefits of the LAG when the data moves off the switch to the NAS making the LAG moot.
And of course as I pointed out earlier, should your switch fail, goodbye connectivity.
To be done correctly from a redundancy standpoint, you'd want minimum two switches stacked together so they behave as one unit. You'd split your LAG with 2 interfaces going to one switch and 2 to the other. Then, should one switch fail, traffic is still flowing onto the second. This is of course ignoring the chokepoint issue.
It matters not how straight the gate,
How charged with punishments the scroll,
I am the master of my fate;
I am the captain of my soul.