I found some interesting disucssion going on on WebHostingTalk.com. I am going to paste some excerpts for your interest. You may follow complete discussion on WHT here http://www.webhostingtal...showthread.php?t=640823.

===================
Here is the Question.. Why would you want MPS? and Why would you Not? MPS has some advantages. I want to clarify, that I have not completely made up my mind that MPS is the best choice. I consider the discussion to do MPS a Chevy/Ford Battle.

Good Things about MPS
HMC/MPS is well documented. Microsoft has built a reference architecture and compiled step by step instructions on deploying this reference architecture. Microsoft had to define a standard of some kind and in most cases the reference architecture is way too big budgetarily. Basically, an experienced HMC Architect can design a smaller solution to fit your needs. The advantage of having MPS is the processes are tested and SUPPORTED by Microsoft. Also, Microsoft has written almost any backend function you need. You just have to write the front-end side.

In MPS if a request fails for some reason, the engine will back out or not perform the process. This is a huge plus. If you were running some script to do a process, most likely, you would be stuck cleaning up a mess.

A quick sidebar... Microsoft expects customers to develop their own front-end provisioning website. The solutions comes with sample code, but it is clearly NOT production ready. You can submit code to the MPS Engine from the command line. Most control panel vendors that are MPS Based (including eJadSPM/HEE), are selling you Front-End websites that interact with the MPS Engine.

I could rattle on all day (too late) about MPS coolness... Frankly, when it boils down to it, most control panels that work with MPS are very high and the vendor take some of the revenue per month per mailbox. Optionally, you can develop your own control panel, but this MPS Developers are a breed themselves.

Disadvantages of MPS
MPS is usually expensive to deploy because you have a high control panel cost or development costs. Then again, if you develop your own code you get exactly what you want. One of the biggest reasons to dislike the HMC Solution is patching and software releases. Example: HMC4.0 has a certain level of Service Pack Approved. If Service Pack 3 for Windows 2003 came out tomorrow, you would void support or even potentially BREAK your environment if you install. You have to wait until the Service Pack is approved for use. Patches between Service Pack's are ok, but I have seen them break MPS before. Another Example, OCS 2007 is not support by MPS. However, customers may want this service or CRM or Other stuff. If you are running MPS, you have to wait until they are slipstreamed into the solution. Non-MPS coders that build control panels can generally fix the code to work with newer service packs or drop new software into the solution faster than MPS.

I know my statements above may appear that I wear an MPS Hat. Please understand that I work with HMC everyday so I know more about it than how people that go direct work their magic.

If you are going to start in Exchange Hosting today, ENSURE the solution you pick is Exchange 2007 -- NOT 2003. A one server install is not a solution that is responsible.

Summary: If you are serious about doing MS Exchange hosting your choice is HMC/MPS architecture.