In a recent note, McKown, John said:

> Date:         Wed, 15 Jun 2005 14:23:40 -0500
> 
> If doing a periodic "T CLOCK=hh.mm.ss" is sufficient, then I'd cheat.
> I'd have a PC which is running a TN3270E emulator. This emulator would
> connect to z/OS as a console (Visara, 2074, ICC, or SMCS console). Have
> this PC sync its clock via the Internet. Have this PC run a script
> periodically which issues the "T CLOCK=hh.mm.ss".
> 
> If you don't want a PC with a console, use a PC that somehow does the
> same thing. But you will need a PC somehow telling something on z/OS to
> issue the above command.
> 
> There is __no__ way to update the TOD clock other than via an ETR.
> 
That z/OS can function as an NTP server, but not as an NTP client
is arrogant, perceived by non-mainframe-partisans as "radically
immature".  Not everyone can be emperor.

That said, what are the practical considerations?  What margin of
inconsistency is tolerable?  If all systems can be synched to an
atomic time standard, by whatever technique, within half that
margin, you have satisfied your requirement.  If z/OS can not be
so synched, then I'll consider the z/OS or z/Series design deficient
for your purpose.  Likewise, of your non-mainframe systems can
not be so synched, they should be considered deficient.  (I'll
ignore any putative requirement to sync all systems to a single
standard that deviates from IAT.)

The ETR is, AFAIK, a PS/2 running OS/2.  It has long mystified me
that that PS/2 can use as a time reference WWV time signals, either
by dialup or by shortwave, or a serial connection that spoofs the
WWV protocol, but not GPS nor NTP.  Why  isn't there an NTP client
for the ETR.

The ETR can be directed to adjust the TOD clock gradually, monotonically,
by operator command.  Does the ETR support a remote console?
if so, that remote console ought to be replaceable by an emulator
which converts the output of NTP protocol to ETR commands, which
would meet the requirement.  The NTP client and console emulator
could run on any suitable hardware, perhaps even the mainframe
itself.

-- gil
-- 
StorageTek
INFORMATION made POWERFUL

----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO
Search the archives at http://bama.ua.edu/archives/ibm-main.html

Reply via email to