[time-nuts] Thunderbolt accuracy...??

[Magnus Danielson]

Lux, James P skrev:
> 
> 
> On 12/20/08 4:44 PM, "Magnus Danielson" <magnus at rubidium.dyndns.org> wrote:
> 
>> Edwin B. Walker skrev:
>>> I wonder if companies don't junk equipment because "electrolytic capacitors
>>> last years not decades" Does this  make sense?
>> No. Details like that does not comes into play. If the quality of the
>> gear does not match the specs, the manufacturer gets a serious problems
>> so they choose components accordingly. They kick the manufacturer around
>> or they choose another if worse comes to worse. Shifting out systems is
>> a much bigger thing. It is about being able to provide the needed
>> services, ensuring the revenue stream. Shifting out old systems can also
>> release frequency spectrum for new more efficient services. This may be
>> at the control of the spectrum authority.
>>
>> So issues like poor capacitors does not really come up. Telco runs quite
>> a different ship than consumer PCs. Way different.
>>
> 
> Even if the basic function is sound, in a large operation, there's value to
> having a limited number of different kinds of equipment around.  You have
> smaller training costs, interoperability is easier, etc.  Say you have 5000
> cell sites, each with a widget that's a few years old, and now you're adding
> another 5000 new sites, and the old widget's not available.  It might
> actually be easier and cheaper in a lifecycle cost sense to buy 10,000 new
> widgets and scrap/junk the 5000 old ones.  Your maintenance manual only
> needs to cover one kind of widget, your depots only have to stock 1 kind of
> spare, etc.

Indeed. When running such a large network, having a consistent network 
management system is crutial. The network management needs to know all 
key boxes in the system, be able to control them, be able to take alarms 
from them and be able to provide good information to the team doing the 
24x7 hours of network operation. These people cost good hours and must 
be able to take actions to ensure the network functionality. Large set 
of diversity is a problem in such cases. Designing network management 
software that makes the two boxes act and behave in the same fasion, 
making the manuals for the support team etc. becomes more expensive. To 
support that telcos spend alot of effort to ensure there is standards 
and multiple vendors for the same box-role to ensure that not a single 
vendor can skirt them up or dry up.

> Furthermore, the "book value" of the widgets being scrapped might be zero
> (having taken accelerated depreciation of 3 years, for instance), so from a
> corporate standpoint, selling them for cheap is a good thing.

Indeed.

One should recall that for a telco, the actual equipment cost is only a 
part of the cost. Operational costs, maintenance cost and installations 
costs includes so much more.

So gear being phased out from a telco may have no value for them but may 
not be without value outside of their framework.

However, not phasing out equipment in time to keep consistency can also 
bite back. One telco for instance had their complete network failed as 
their GPS receivers (same model thoughout) did not support a certain PRN 
number, and as long as it was not overhead they would work, but fail as 
it came up again.

Let's just say that the customers where not impressed and happy about 
it, they had to make a very quick (and hence more expensive) buy-up and 
replacement.

> You'll see this in the desktop PC market for large companies, where the cost
> to support multiple hardware configurations is substantial, and where
> there's a fairly standard 3 year amortization cycle.  At some point, keeping
> that 5 year old PC working just isn't worth it. (Oh, only Bob knows how to
> fix that one, so you won't get your computer fixed today, because Bob's on
> another call or on vacation, sorry, key employee doing mission critical job,
> you're out of luck.)

Another aspect may be that you want to shift them out before they die on 
you and in a way that kills you more than the loss of the machine... 
loss of data. Loss of access to data is annoying but can painstaking and 
critical in itself, however usually that spreads out a bit. Loosing 
critical data and not being able to get it back is worse.
Being able to support the set of applications needed and upgrading OS 
and applications is an issue. In just about three years machines seems 
small and weak, but can still work on for a while. In 5 years another 
upgrade becomes troublesome. At the same time the actual hardware cost 
as gone down. Upgrading to a OK machine has become much less a hardware 
issue than the cost of managing it.

> Even for test equipment, there's good reasons to cycle through new gear in a
> fairly short time. Especially if you're changing test configurations a lot,
> you want to encourage "model number independence" and not depend on
> peculiarities of a particular instrument, otherwise you wind up with
> specialized test sets with 20-25 year old signal generators (I'm looking at
> YOU, you old HP 8663s) that can't be repaired, and because nobody has been
> following along, using something new is a major jump, rather than a small
> update. (And you find out that Agilent Exxxx models have very different
> operating peculiarities from their predecessors with similar model #s, even
> if they do work better by all rational evaluations.).

Developing and revising test rigs comes at a fairly high cost. Being 
able to support the same command-set to ease drop-in-replacements is one 
way to achieve things. For a production line setup, being able to swap 
in instruments to handle scheduled calibrations should also be 
considered. As a source of instrument dry up, this could lead to stalls 
in the production. In the end, it could mean the death of that product.
With dry up it can mean that the instrument support has been dropped for 
instance. Being able to keep it calibrated could be an issue. Loosing 
calibration due to loss of battery power is another issue.

> To those tinkering at home, though, this cycling through is great.. If
> you're willing to fix it yourself, and maybe have a hangar queen or princess
> for parts, or you don't need ALL the functions to work (never needed that
> knob anyway..), then you too get really nice gear to work with. (and,
> besides there's something really satisfying about that orange glow of the
> Nixies... It impresses people who see your garage, because its redolent of
> Atomic Age science fiction movies of the 60s and 70s.)

Certainly. Let's not complain or whine about it. We are winners on this. 
Most of the time.

Cheers,
Magnus