Mainframe clock explantion
     For lack of very expensive Sysplex Timer hardware which is otherwise unnecessary in our environment, the mainframe Operating System (currently z/OS V1R7) uses an internal hardware clock for multiple purposes beyond just logging events. It cannot tolerate duplicate times; i.e.; a particular mmddyyyyhhmmsstthh must not occur more than once. This requires resetting the hardware clock when going to and from Daylight Saving Time, which impacts us most severely in the Autumn when we fall back an hour.

     Resetting the hardware clock requires that the Operating Systems be shut down on each LPAR and the hardware undergo a Power-On-Reset, after which z/OS is IPL'd (booted) on each LPAR. The POR is what puts the new clocktime into effect for z/OS to pick up at IPL time.

Shutdown takes about 10 minutes, POR/IPL about 30 minutes.

Shutdown/POR/IPL done without time change:
01:10am - z/OS Shutdown
01:20am - POR/IPL started
01:50am - z/OS back up
User-experienced outage = 00:40   (Shutdown + POR + IPL)
Systems-recorded outage = 00:40

Shutdown/POR/IPL done in Spring to go to Daylight Saving Time:
00:05am EST - Clock advanced to 01:05am
01:10am EST - z/OS Shutdown
01:20am EST - POR/IPL started
02:50am EDT - z/OS back up
User-experienced outage = 00:40   (Shutdown + POR + IPL)
Systems-recorded outage = 01:40

Shutdown/POR/IPL done in Autumn to go back to Standard Time:
01:05am EDT - Clock set back to 12:05am
01:10am EDT - z/OS Shutdown
01:21am EDT - POR - clock now says about 12:28am
Twiddle thumbs 1 hour while real time catches up to hardware clock time.
01:29am EST - IPL Started
01:59am EST - z/OS back up
User-experienced outage = 01:49   (Shutdown + POR + IPL + thumb twiddle)
Systems-recorded outage = 00:49