I agree with the statement that you need to look at how your monitoring solution integrates with the rest of your infrastructure. You want to limit the number of places where you are getting reporting information. You don't want to have software tools monitoring at the OS level, the application level, the hypervisor level, and the storage level and the networking level and then have to manually go to each one and try to integrate the findings. You want your monitoring solution to have visibility across the entire datacenter.
Also, you want to think about lock-in. If you invest a lot in one vendor or product, you want to make sure you are going to be OK with the limitations of that product. You may end up wanting one feature in your environment that one product doesn't have. Do your homework to research everything you are going to need long-term - capacity planning, monitoring, provisioning, security compliance, automation, chargeback/showback - and what your solutions do at the application, OS, hypervisor, networking and storage layers as necessary. Design a solution that is going to allow you to integrate as much vision of the environment into one place as possible, and make sure you are ready to live with that product for a while.
