How does a 2D MEMS based optical cross-connect control its mirrors? Concerning fiber optics, I'd like to know how a 2D MEMS based optical cross-connect is able to switch optical signals. I've read that these mirrors are controlled with electrical signals to be either "on" or "off" to redirect light to output ports. 
As far as I know, all-optical based switches don't use buffers on the incoming data. However, an electrical signal must be generated to control the MEMS mirrors in order to redirect the signal to the output port. How can this electrical signal be fast enough to switch the optical signal? I assume this electrical signal comes from the header of the packet, but I'm not sure how an optical switch obtains that information and generates an electrical signal.
 A: First, I would like to refer to a very important fact, which is, many researchers like to grab attention to their work even, by using wrong names. What I am trying to say is that, you will find many work with the "all-optical" word in it, while it is electro-optical. So, you need to carefully read the work to understand the whole operation. A book mentioned this issue and how there is a huge misunderstanding and abuse of terms is the well-known book "Optical Switching" by Tarek Al-Bawab.
Second, I haven't heard about an all-optical MEMS switch. An all-optical switch needs to use the optical input to trigger the switching elements and get directed to the desired output. In a 2D MEMS switch, a mirror is either in the stand position or the laying position, to either reflect a beam, or pass it, respectively. So, I do not think a 2D MEMS switch is capable of doing this.
A good website that might help you understand the idea behind the all-optical switching, and the trends in creating an all-optical switch is provided by a lab at Duke University:
http://www.phy.duke.edu/~qelectron/proj/switch/intro.php
I think it might give you a basic idea of the concept and what they are trying to do.
