[time-nuts] Beaglebone NTP server

Chris Albertson albertson.chris at gmail.com
Fri Dec 12 14:42:11 UTC 2014


Let's say you had followed this advice in 2004 or 1999.   You build a
single purpose NTP server out of whatever was the best smaller server
platform back in 1999 and today it broke.
Even if you could, would you want to replace it with a 1999 vintage small
server computer?  No way.  You'd notice the BBB as being better, lower
priced and using less power.

So I assume that in 2029 when the BBB fails after 15 years there is no way
you even want anther BBB.  For one thing the BBB would not run any OS that
you were foamier with.  It would not run the current best version of the
software and.

Actually my 1999 vintage small server was replaced BEFORE it failed because
it was burning about $8 in electric power per month.  That adds up over a
10 year period.  It got replaced with something that uses a tiny fraction
of that power and paid for itself.

On Thu, Dec 11, 2014 at 1:23 PM, Hal Murray <hmurray at megapathdsl.net> wrote:

>
> jimlux at earthlink.net said:
> > Ah, but will the exact same single board computer be available for
> > replacement in 5 years?  Or will it be Rev F instead of Rev B, with
> "just a
> > few tweaks to improve performance", but also enough that it's  not "drop
> the
> > image on it and run"
>
> > What about 10 years? 15?
>
> My straw man is to buy a bunch now and put them on the shelf until you need
> them?  It would be interesting to read a good B-school type report on that
> topic.
>
> Is it realistic to expect any electronics related stuff to be available for
> 10 years?  I think I've seen a few boards marked long-lifetime.  I don't
> remember the details.  At least one OS distro has a version marked
> long-lifetime, but I don't know how long that covers.
>
> What does the military do for this problem?  What's the average lifetime of
> something like a radio in a military plane?  How often do they get upgraded
> because the old ones are no longer maintainable due to unavailability of
> parts?
>
> What's the average lifetime of a fab line?  Or even the max lifetime?  (in
> case you are a big vendor with multiple fab lines for a particular
> technology)
>
> What's the typical lifetime of parts used in timing related gear?
>
>
> > Maybe that's the key.. think of it as an appliance. If it stops working,
> > and it's not because of something simple (power cord), then you're
> probably
> > better off building a new one from scratch than trying to fix  the old
> one.
>
> > That is, the labor involved in "port to a new platform" might be
> > substantially less than "find old platform and install it in old box",
> if
> > only because things like tool chains tend to follow the latest hardware.
>
> A lot of that depends on the quality of the documentation you kept when you
> set things up.
>
> I've worked with guys who were good at making a big pile of kludgy gear
> jump
> through hoops.  One of them was very good at writing down what he did.  It
> doesn't have to be a fancy document blessed by 43 proofreaders, but it
> really
> helps to mention all the critical steps and/or explain the non-obvious
> reasons for doing things.
>
> I've gotten into the habit of making a checklist of what I have to do to
> install distro-X from CD and fix it up to do what I want.  Every now and
> then, I want another box just like that one except...
>
> It isn't fancy, but it does have a line to remind me about each file that
> needs editing and what tools are used to set things up.  Often there are
> chunks of code I can cut-paste.
>
>
> > Even if you kept the tool chain for your old platform, running it on a
> new
> > computer might be problematic. (recognizing that there are people  out
> there
> > running IBM 1401 emulators on System/360 emulators on... but  that takes
> > time too)
>
> It's not uncommon to archive the PC that has all the tools when a
> significant
> hardware project is released.  Just put a big note on it and push it into a
> closet in case you ever need it again.  (Setting up CAD tools is often
> quirky
> and hard to debug.)  Again, this needs documentation.  Suppose you want to
> make a simple one line change to fix a typo.  What do you do to turn the
> crank and make the files you hand off to the next step?
>
>
>
>
> --
> These are my opinions.  I hate spam.
>
>
>
> _______________________________________________
> time-nuts mailing list -- time-nuts at febo.com
> To unsubscribe, go to
> https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
> and follow the instructions there.
>



-- 

Chris Albertson
Redondo Beach, California



More information about the Time-nuts_lists.febo.com mailing list