Re: Grand Apagon - Electricity (not) in Spain

On 08/05/2025 12:49, Don Y wrote:

But where can exchange traffic go?  See what I mean?  Anyone that
you want to contact (and everyone along the way) must be "up".

>
That was the original point of ARPANET then EPSS and later the internet. Packet switching means that any route to the destination at all will do.

 But that assumes there *is* a series of hops that can get you "there"...
wherever "there" happens to be.  In a nationwide outage, what chance
that everything EXCEPT some critical bit of comms gear is affected?

Nationwide outages should be exceptionally rare. Never had one in the UK but the way they are going one winter's day it will happen.

I'm told that my fibre feed is passive optical connectors and splices all the way back the regional exchange about 12 miles away. My local exchange was about 5 miles away and a so-called exchange only direct line (which meant that ADSL 2+ was the limit for me prior to FTTP).

 So, you rely on the exchange having upstream connectivity.  Along
with the fiber link TO the exchange.

That is usually a given since the regional and further up the chain concentrators all have UPS and diesel  generators or fuel cell supply.

My mobile phone worked all the day, I could send and receive whatsapp messages.

>
Are those processed "locally"?

>
Mobile phone masts here typically have a lifetime of about 8-40 hours after power failure depending on how heavily they are being used. Backhaul presumably is optical or microwave.

 So, also subject to outage.

But the central nodes usually have better battery backup and/or generators than the local nodes. Local nodes die first according to how much traffic they have to handle.

Most powercuts tend to be fairly local round here - a regional powercut or a national one requires something truly catastrophic to happen.
>
I can only recall one UK powercut in that league in the past half century (August 9 2019). Of course it directly affected the densely populated affluent regions London and the South East. Therefore it was much more newsworthy than if it had affected the remote Scottish Highlands where weather induced powercuts are quite common.

 I don't think I've ever (regardless of where I've lived) experienced
a deliberate power cut.  A drunk may take out a telephone pole or
a branch may fall on some high tension wires but no one has ever
said "sorry, we're turning the lights out" (for whatever reason)

They do that routinely where I live once a year for trimming trees that might otherwise short out live supply lines or worse fall onto them.
I wouldn't describe that 2019 powercut as deliberate either it was a huge MFU caused by a single lightning strike to an insignificant power plant that lead to a cascade network failure.
We typically lose power a couple of times a year due to very high winds toppling poles and/or the sort of snow that sticks onto trees and makes them break. The mains poles here are now antique. Installed ~1950's and the bases are rotten. Unsafe for linesmen to climb and marked as such.

The recent big one at Heathrow didn't affect all that many people although it did take down the whole airport which shows remarkably bad contingency planning - it should have had supply redundancy and the ability to switchover to it before the diesel generators ran out of fuel. Heads should roll over them having to shut down completely.

 Fukishima?

Genuine natural disaster beyond what the designers had considered. They almost got away with it but didn't. Tsunami are absolutely terrifying. Another one when I was in Japan in the hours of darkness was reported as  >2m (which was when the gauge stopped transmitting). The next morning there was seaweed hanging off supergrid pylon wires.

I'm considering replacing the UPS at my router. Some UPS "destroy" the battery too fast.

>
Yes.  Rather than spend time investigating it, I've taken the approach
of just rescuing batteries to replace those that have been "cooked".

>
That is a feature of UPS design that specsmanship to get the longest run time for the sales datasheet means that they cook their batteries. I have seen them swell to the point of bursting inside a UPS. Thick rubber gloves needed to remove the remains. Support metalwork was a real corroded rusty mess but electronics above it remained OK.

 Yup.  They have a rationalization, though -- they are trying to provide the
highest availability.  Else, how much availability do you sacrifice to
maximize battery life?  Do you then start specifying battery life as a
primary selection criteria?

I think they probably could back off the fast recharge a bit. I'm always nervous of going back on again too soon after power is restored (even though my systems are reasonably fault tolerant). Sometimes the mains restoration goes on and off several times a few seconds apart if there are still other transient leak to ground faults on the lines.

[Most SOHO users buy a UPS -- thinking they are being "professional" -- and
then discard it when the battery needs replacing and they discover the
costs charged by the UPS manufacturer -- or local "battery stores"]
 

I suspect the problem (rationalized by the manufacturers) is trying to
bring the battery back to full charge ASAP -- as well as keeping the
highest state of charge that the battery can support.

>
Which taken to extremes is very bad for battery life.

 Of course.  But, they are in the PRIMARY business of selling batteries,
not UPSs!

A bit like printers then.

Charging at a slower rate and to a lower float voltage would
compromise the UPS's availability -- but provide less maintenance costs
(of course, the manufacturer wants to sell you batteries, so you
can see where their priorities will lie!)

>
They really think I'm going to buy their vastly overpriced replacements?

 If you were a business, it would just be a maintenance expense.
You would budget for it.  If SOHO, you'd likely replace it at
most once and then realize "Gee, I haven't NEEDED this in the
past three years so why am I spending more money on it?"
 With the exception of multi-user servers, individual workstations
usually have auto-backup provisions *in* the key applications.
And, in the event of an outage (even if the machine stays up),
the user is usually distracted by the rest of the house/office
going black; is ~15 minutes of uptime going to be enough if the
user isn't AT the machine when power fails?
 No one has yet to address the market where TCO is the driving
criteria.

To some extent it is an insurance policy to not lose what I'm working on if the power does go down suddenly. Despite having theoretical lightning protect as well I also shutdown when there are thunderstorms about.
I saw what a big lightning strike to our works building did to the switchboard and mainframe. The surge protection devices on a big chunky copper bus bar saved themselves by allowing transients to fry all of the terminal driver boards. The phone lines were just a sooty shadow on the wall and it blew the clip on covers off the cable way.
About once a decade we get lightning to tree strikes within 100m. It usually fries bedside clocks and modems (although mine survived OK last time). This was despite a 1" calorific spark jumping off it.
--
Martin Brown