Has anyone investigated the possibility of being infeasible to shut down power on FDC's or RDC's (panel board distribution) feeding multiple server racks? If you have a production data center environment that supports many sectors of your core business. Is this justifiable to perform live terminations 120v/208v 30amp inside of power panels by means of qualified person?

What do you do when there is a power failure?

By "Live terminations" do you mean breaker replacement/installation? New wire installation (conduit/cable entries, wire pulling, etc)?

I thought a good data center design included redundancy down to the server level. This means working live and ignoring the possibility of an unintended shutdown is a "culture" issue that needs to be changed.

I have not had a power failure yet!

Yes, I am referring to breaker installations, new wire installations (conduit/cable entries, wire pulling, etc)

Are you referring to working live from an electrical installation standpoint?

What are your plans in the event of a power failure, do you have built in redundancy to enable you to shut down one area while keeping the others up and running?
There is no reason beyond what NFPA-70E and OSHA allows to work on any circuit while it is energized. What would you do, besides being injured or killed, if you accidently caused an arc flash incident. How would your system respond and how will you be able to restore it, if possible. If you can't answer these questions, you should start developing procedures to deal with the eventuality of a power failure for what ever reason. If you system does not have any redundancy and your business requires it to be up, I suggest you get together with your engineering group and figure out how to accomplish deenergizing part of your system while keeping the remain part working.

I agree with Rich. I have done plenty of shutdown jobs at data centers, there is always a way to work around working live.

Wear the PPE that is appropriate for the rating of the panel. If the PPE dimishes your dexterity to the point you can't safely work, then it will be your call to plan an outtage and complete the work..
IMO..

The wearing of PPE should not be based on dexterity, it should only be used as a last resort. Designing a system to deenergized it in sections so that the remainder can still function is the correct way to approach this problem. OSHA and NFPA-70E allow working on a circuit energized only when you are doing the following:
1. Troubleshooting
2. Restoring Equipment to Service
3. When it creates more of a hazard then leaving it energized
OSHA allows a fourth, when you turn out all the lights, but with 70E removing that in 2009 you would be hard pressed to explain that to OSHA, as to why you couldn't use portable lights when working.

I would not determine what PPE to use based on whether 'it fits' or not.. What I meant, and have used in the past as a point on why to shutdown the equipment, was if the PPE (specifically the gloves) are too cumbersome to work safely in a panel live, then a short outtage would be required while you complete that portion of the job. Some of the older equipment has not been designed with lots of room to work in safely live...

How would you have a lighting panel designed with a split bus to allow half of it to be shut off with the rest of it energized?

The split bus would not solve the problem of working deenergized, as half the panel would be alive. The better fix would be, if the lights had to be on, would be to put half the lights in one panel and the other half in another. If you were working on individual lights, changing ballasts, then install the quick disconnects on each light.
The point that I am trying to make is that you shouldn't be working in a lighting panel at all while it is energized to do other than the 3 items I indicated that are permitted by 70E. Then, you don't have to worry about wearing cumbersome gloves. Although it is definitely less cumbersome then being in a burn unit if you have an arc flash.

If all of the rack equipment had dual corded power supplies, "A" and "B", powered from "A" and "B" PDU's, then either side could be de-energized without causing an outage at the racks. Sometimes the redundancy goes all the way back to the UPS, with a seperate unit for each side.

This is probably more for redundancy than for safety, but it also permits work to be performed de-energized.