LACP and LBT and two different approaches to a similarly-hopeful result. As such, they are really mutually exclusive. When creating a LAG, one bonds one or more uplinks into a logical bundle. The LAG algorithm handles distribution of traffic amongst the links. LBT does so similarly although in a more intelligent way and without the need for any upstream switch configuration. So if you use LACP, there's no point in using LBT. When using vSphere, there is a debate between use of the two. Networking guys usually want LACP because it gives them control over the networking aspect and may simplify switch configuration. But LBT is, at least in my opinion, the better choice in a vSphere environment because, unlike LACP, it has the ability to load balance the links in a more even fashion because the algorithm looks at physical link saturation levels, something that LACP does not. While LACP distributes load with the hopes of it being equal (it is almost never), LBT has more intelligence built in to the point where, if any link becomes more than 75% saturated, it will move traffic over to another link, effectively balancing (not distributing) load.
