Posts by SImon Hobson 2539 publicly visible posts • joined 9 Sep 2006
I cut my teeth (as it were) on VAX VMS. Even today the DCL command line environment from VMS seems futuristic compared Linux, Windows and anything else I have looked at.
Funny to think back at how the Windows lot used to laugh at us Linux type and all that stupid and hard to learn command line stuff we had to do.
And now Windows is all about Powershell and complicated and hard to learn command line stuff.
Really ?
The article named two of them, and there aren't that many cloud "giants".
Microsoft - accused of charging more to let you use their software on any cloud but Azure.
Oracle - long history of screwing over the customervictim with questionable licensing terms.
Of the "big guys", I think that leaves Amazon and Google - do I really need to say more about them ?
And it's true that from my understanding it's illegal to use any of these to hold personal information - though I assume all of them have good enough stories to placate most businesses enough to believe that they can. For example, Microsoft makes a big thing of you being able to decide where you data is held - but the dat the CLOUD act was passed, Microsoft in the US handed over data held on servers in Ireland which rather proved that their claimed access controls didn't actually control that illegal access.
Unfortunately we've seen wha MS does when governments require compliance to open standards - they'll stuff the national standards bodies with shills to vote through a piss-poor pile of manure with a deliberately confusing name becomes an "open standard that no-one but MS can implement", where conformance (or lack of) is impossible to demonstrate, and which simply carries on the "Use MS or have everyone complaining about lack of compatibility" control over business computing. With Microsoft, it seems that when they use "open" in the title (as in Office Open XML for the 'open' standard), it's like when countries have "democratic" in their name - you take it as an indicator that it's anything but open or democratic.
Now, if governments added that the standards have to have a compliance verification tool available which is also open, and alternative implementations, then things would be different - MS would need to write a competing pile of manure in order to be able to tick the box.
Yes, but ...
The problem isn't that we have 14 different standards, it's that we have one de-facto standard that everyone except MS uses - because to MS, a standard it doesn't control in order to prevent competition is something really bad. All we need is for that one bunch of ****ing ****s to support the existing standard that others already support - no new standard required.
I don't think you realise the situation.
This isn't like deciding one dominant vendor is bad and chucking them out - this is a natural monopoly as there can't really be competing registries for the same part of the internet name space. Well, given enough effort you could have two (or more) - but they'd need to be somewhat linked to avoid having the same name registered by both and in reality it wouldn't be likely to work well.
But what it does need is for the de-facto monopoly holder to operate in a transparent way, charging only what's needed to pay the bills and keep a bit in reserve to cover any hard times - and with those bills (including staff and director pay) kept reasonable. Given that Nominet already has all the tools and processes in place, it would be foolhardy to toss all that out with the bathwater and start again. Far better to fix the fairly simple problems - kick out the pork troughers, stop dabbling in dodgy investments, and carry on running the registry as it should be run.
They're doing it in Bristol - and like the other example given, causing massive disruption while doing it.
But you are right - it's a properly joined up way fo doing things.
The big problem is likely to be the tiny little issue that they tend to put nuclear reactors a long way from lots of houses - something about people not wanting them in their back yards ? And district heating does work best when your source of heat, and the use for that heat, are reasonably close together.
I'd add that without government being on-side and supporting it - by policy, not necessarily funding - it would be a massive risk. Just think if you've just invested [insert large value here] and then the government turns round and bans nuclear as happened in Germany as a knee jerk reaction driven by the greens after Fukashima. I suspect there's been more radioactivity released from the extra coal burned (but it's OK, that happened in Poland so doesn't count) to replace the nuclear.
And over here, we've had a bit of a roller coaster ride in terms of political support for nuclear. Traditionally Labour have been agin it, while the Tories have been "sort of supportive but frightened of the minorities who shout very loudly". But as you point out, there's nothing like a good crisis to persuade people to stop kicking that tin can down the road.
Actually, it's not theoretic as AIUI it has been demonstrated with a real world operating system. In theory, it could take most of what we currently label as "waste" and reduce the actual waste bit to a fraction of what we are currently paying to deal with. AIUI there was a proposal from someone to build one near Sellafield but for whatever reasons it didn't happen - and I suspect that in part objections to anything creating (even temporarily, before consuming it too) plutonium.
If we did build a few fast breeders, I believe we have enough "waste" in storage to supply our lecky needs for a century without digging any more uranium out of the ground.
It does seem crazy. We have a supply of fuel, we have the technology, but instead of using that fuel we are going to spend a fortune disposing of it because some [make up your own description here] prefer to label it as waste.
Yes, the mess left behind by past endeavours is ... a bit of a mess. But these days, what to do with things at the end of life is considered during design.
Leaving aside some of the "less clever" ideas from the past, you also need to be aware that some of our waste problems are actually cased (in part) by demands from the anti-nuclear lobby. Things like dismantling reactors "hot" instead of allowing them to cool down (radioactivity wise), and treating what would in any other circumstance be "mostly fuel, just needs processing a bit" as waste that must be expensively disposed of.
A big difference is that the technology is mostly already there, AND you don't get to have decades of civils needed.
The reactor itself will be a factory built unit that gets put on a lorry and delivered to site already fuelled and ready to go. Slot it into the support systems built on site and away it goes. As the article mentions, the idea is that there will be many of the same design (like the French more or less go to with their big reactors) rather than lots of different and bespoke system like we were daft enough to do.
But a massive part of the cost saving will be down to the support and safety systems. Our current big reactors need lots of safety kit to keep them working and safe - we saw from Fukashima what can happen if you take away those support system while there's a lot of decay heat still in the core. Other than the Westinghouse AP1000 design which incorporates truly passive cooling for a day or two while you get some support restored, they all need these systems which are very expensive - and hence drive the push for ever bigger plants.
With the SMRs, they are intrinsically or inherently safe - basically they are designed in such a way that you could remove all the support systems, the reactor will shut down, and passive systems will cool it enough to prevent a meltdown or nuclear material release. The reactor may or may not be useable afterwards, but there won't be contamination, or exclusion zones, or hydrogen explosions (which made for good, but innacurate, TV at Fukashima). Just lift the reactor out and ship it back to the factory, fix the problems, pop a new reactor in the slot.
That's where most of the savings will come from, by not needing the massively engineered safety systems, and huge crowds of very highly paid people to look after it.
Your kettle may be 2 to 3 kW - but it's only on for a short time. The average, across many houses, is actually quite low. At a local level I believe the DNOs work on around 1kW/house for sizing the local network.
On a national level, the total load tends to be in the order of 30GW, rising in winter. So with something in the order of 30M homes, that would work out at 1kW/home if you ignore all the industrial users.
But as you say, with the drive to electrify everything, that average will be going up considerably.
I'm not.
You can't expect a business who's business model depends on ignoring privacy laws to interpret the laws in a way that suits itself. And then it takes some time for an interested party to gather enough evidence to persuade a national data protection body to take action. And then it takes time for that action to happen.
Compared with the response Shrems got from the Irish data protection people, this one seems to have happened quite fast.
Not this old lie again.
The SOGAS Act does NOT give you 6 years warranty. It says that goods must be as described, reasonably durable, and free from defects at the point of sale. It does not make any statement at all about what timescale would be "reasonable" - that entirely depends on the situation (e.g. paying a lot of money for a high end device has a longer expectation than some cheap knock-off, and it may even be longer than 6 years. Personally, I don't think anyone with any knowledge of electronics & IT would consider 6 years a "reasonable" expectation of fault free performance - even for Apple - although it's common for stuff to last a lot longer (I've just upgraded from a 16 year old Macbook Pro to a 6 year old one, but only because I needed more up to date software support).
The 6 years (5 years in Scotland) comes from a completely different piece of legislation - statute of limitations. That basically stops you taking civil action for a breach after 6 (5) years. It places an upper limit on your consumer rights, it does not mean you have a "6 year warranty".
That Apple page is very misleading and just inviting problems for themselves.
My guess, they simply declared "third party hardware at fault" and the $99 fee is to remove that and put a "properly working" piece in. In fact, that's pretty well stated in the article - the service centre removed the user's SSD and fitted a different one which allowed the firmware upgrade to proceed.
The OS, drivers and software that ran on the older kit will continue to do so. If you want something he older kit didn't support then you will need something newer although the older products might need continuing security releases.
Hang on a minute there.
I have an old iPad, it still does everything it did when I got it ... except for the functions Apple nobbled. I can no longer transfer stuff like films (what I used to use it for most, when travelling) because they took the USB sync feature out of OS X. And they have a policy of not keeping anything but the latest version of any software around on the store - so while there would have been software that will run on it, Apple actively blocks me from downloading it. I'm no stranger to running older software on older hardware - but on typical desktop OSs it's not generally hard getting hold of older versions, but in the Apple iThing world it's actively blocked.
Comparison of the source and executable of the compiled program should ...
Except that it doesn't. The days of a "dumb" compiler that would produce predictable code from a piece of source are long gone. These days, the code is heavily optimised to match the target processor - so even very minor changes to compile conditions can produce significantly different code.
AIUI it's one of those areas that's had a lot of attention over the years.
Personally I would have chosen M$ as they have history of fighting overreach from the Cloud Act. But the same arguments apply about AWS.
AWS UK will be operating as UK entity and the ability of the US Govt to compel that entity is limited in exactly the same way M$‘s US entity was unable to compel its Irish entity.
Err, would this be the same MS that the day the CLOUD act was passed, just handed over the data, housed in a datacentre in Ireland and operated by the Irish subsidiary of MS, some US TLA had been after for a couple of years ?
Thus proving beyond a doubt that having a datacentre in the UE, subject to EU law, and operated by a supposedly legally different entity, and supposedly having technical measures in place to prevent the US parent company from accessing the data ... did didly squat to prevent staff in the US from accessing that data and handing it over.
As it is, I've seen enough of the way MS handles logins (a very long chain of redirects, most using domain names under the control of the US company, to suggest that claims about territorial security of data are ... "a bit questionable".
I suspect it's actually quite hard to setup an arrangement where the data is located in the UK, is under UK control, and there is no legal or technical mechanism for the US based parent company to grab or demand access to the data when instructed to by a US TLA.
Like Standard Oil, and IBM, and ...
And with the particular example quoted, the problem is that Google isn't like the supermarket. It's like a megalith that not only runs the supermarket, but also the local radio, and local TV, and local papers - and it deliberately nobbles Mom & Pop's attempts to advertise - effectively delaying each of their ads for long enough to put their own up first.
And of course, we've not even started to get into territory like SO and IBM. In both cases they would target a market segment in an area and actively set out to cripple competition. In SO's case it was petrol etc, in IBM's case it was (originally) cash registers (and especially independent repair/overhaul services). The standard tactic was to open up in the area and heavily discount products below the cost price of the incumbents, or offer silly trade-in prices (making the sale of a new one below cost price) for your old cash register so they can take them out of the market. The existing second hand cash register dealer goes bust because they can't get used ones to refurb, and can't sell what they do have; the incumbent petrol retailers go bust because they can't compete with SO's below cost prices. Then once the competition is gone, up go the prices (massively) to generate the profits for the next targeted campaign.
AIUI both were well demonstrated, SO was found guilty, but IBM miraculously had the investigation dropped on a change of government - absolutely nothing whatsoever to do with their campaign funding !
So, is the supermarket selling below cost - or just has a much better supply chain ?
I would agree about the right to not join one, but the rest of your post is a load of bull.
It's been demonstrated time and time again that within Amazon any "discussion" around pay, conditions, etc is basically a case of "back to work and if you raise it again you're fired" - if it isn't just "you're fired" as an opening response from the company. There is a MASSIVE disparity in bargaining power between Amazon and the warehouse workers - with Amazon basically able to ditch anyone not prepared to work like a slave, under unreasonable conditions, for a pittance in pay. So having a union with a mandate to negotiate on the workers behalf is unlikely to be anything but a good thing for the workers.
And without investigating, I would suspect that Amazon have set up shop in places where there isn't too much competition for workers. That's a common tactic - offer jobs where there weren't too many and people will welcome you with open arms. But then they find that the employer is a [insert expletive here] and unfortunately there aren't a lot of other options.
Ah, that sounds like a "lost neutral" which is a known problem with TN-C-S (a.k.a. PME) supply systems.
Basic lecky supply description. There's a 4 core cable runs down the street - three phases and once combined neutral and earth (CNE). The CNE is earthed wherever it's convenient (typically at joints) and the DNO (Distribution Network Operator) will have rules about how many and how "good" these earths must be.
A typical domestic customer gets a two core cable* which brings in the CNE and one of the three phases. At the services head (a.k.a. "main fuse"), the CNE is split to provide both your neutral connection and the earth. The more awake will already have spotted that if the CNE is broken, then there's nothing to carry away the lecky that's come through the live core and through your appliances - hence the CNE at the user's premises will rise to a significant voltage (could be as much as 240V (nominal) above ground).
But if the shared CNE in the street cable is broken, then multiple houses will be sharing the "not a neutral any more" core, and as the houses will have different loads and be on different phases, the voltage seen between live and "not a neutral any more" will vary between "not a lot" and "up to 415V". This will fry a lot of equipment.
But the voltage between the consumer's earth terminal and ground won't go above 240V (well actually, the upper limit is 253V) relative to local ground. The problem here being that all that carefully "earthed" metalwork - like the body of the washing machine, probably your plumbing, etc, etc - is now "live" relative to local earth.
In many ways TN-C-S is very far from ideal, so much so that there was a proposal (that didn't make it into the latest edition) for the wiring regs (BS7671) to require that every property be fitted with it's own earth rod just to deal with the "broken CNE problem". And it's also why almost all home car charging points now actually have a system to detect this and disconnect the "earth" from the car (after disconnecting the power) to avoid having a car with live bodywork surrounded (potentially) by earthed metalwork like fence posts or lamp posts.
Not, strictly speaking, "inside" the old cast pipe - more, along the old route and surrounded by bits of broken cast pipe.
They have a mole they can pull through the old pipe which splits it and expands it, while towing the new plastic pipe behind in behind. Then they have to dig a hole at every tap off point to re-connect each customer. If the customers are lucky then the same technique will be used to put the new plastic service pipe in - otherwise they'll have their nice smooth drive "re-designed" with a trench.
I think you've missed the point - there is no free ride being asked for.
Apple sold the iThing and made a profit on that sale - if not then that's their own fault. Now they want to make a profit on everything the end user does with that iThing - like buying a game.
Crap car analogy. You buy a car, but then find you can only buy fuel that the manufacturer approves of, and the manufacturer takes a 30% cut of any fuel you put in. Or you can only fit tyres the manufacturer takes a 30% cut of, or ...
Both users and Epic would be perfectly happy to not use Apple's servers at all. The user simply wants to buy and install a game on the hardware they have bought and paid for. Epic simply wants to sell the game to the user, for the user to run on the hardware that the user had bought and paid for.
Apple have artificially rigged the system so that the user hasn't really bought the iThing - they have paid for it as though they have bought it, but Apple have reserved unto Apple the right to determine what the user can do with it, and have reserved unto themselves the right to shakedown anyone wanting to provide users with software to run on their iThing. This is an entirely artificial arrangement, it would only be a free ride IF Epic were wanting to use Apple's services but not pay. But they don't, they'd be perfectly happy to not use Apple's services at all.
Actually, Epic have argued that they'd be happy to not have to use Apple's services at all.
It's an artificial construct - using Apple's services to get their apps onto iThings is something that Apple forces on devs. I suspect that were it optional then we'd not be having this discussion at all - some devs (like Epic) would do their own thing and not "use a free ride", while other smaller devs would decide that the convenience was worth giving Apple 30% for.
Lets say you have a perfectly functional pair of feet, and are happy walking home from the pub. A taxi company sets and forcibly abducts anyone trying to walk home - and charges them for the ride. Some people complain about the fares, are clear that they didn't want the ride anyway - but are accused of being out for a free ride because they don't want to pay the fare.
Well, assuming you mean that everything would end with a quick bang, then you'd be wrong.
Depending on altitude, it takes a while to lose consciousness due to the low air pressure - from memory it's around 15 seconds at 30-35k feet, and the effect tails off so that between something like 10 and 20k feet it will depend to a certain extent on your physique (smokers fare badly here) whether you lose consciousness at all. And if the oxygen masks drop, and people use them, then they may remain conscious even at 30-35k feet.
And after the bang, it could take a few minutes for the big pieces of plane to reach the ground - obviously dependent on what the damage is and whether it "sort of" flies or falls like a stone. So quite likely that many or most of the passengers will have at least a few minutes to contemplate the situation.
In the case of the JAL 747 where the rear pressure dome blew out, the aircraft carried on flying for some time before it flew into the side of the mountain. Reports from the few survivors indicate that quite a lot more probably survived the crash but died due to their injuries and/or hypothermia before rescue attempts were started many hours later the following morning.
Not a nice thought really, but there isn't any truth in a suggestion that when the bomb goes BANG then no-one will know anything about it.
The pilots were faced with what to all intents looked like a trim runaway. They could not have used the "easy to use trim switch" to deal with that because they would have pulled the circuit breaker for the electric trim. The thing is, by the time they would have worked out to do that, the forces involved would already have reached very high levels - and I believe it's a LOT of turns of the manual trim needed.
So there they are, both of them using all their strength to try and overcome the out of trim forces, and they have to manually wind a couple of hundred turns of the manual trim wheel.
Incidentally, had MCAS been properly documented then pulling the electric trim breaker would be the way to disable MCAS - physically prevent it running the electric trim motor. So the same problem - the nose is down, very large forces needed to override the out of trim controls, airspeed and forces rapidly increasing, and you have to wind the manual trim wheel a couple of hundred turns before you either hit the ground or rip the wings off.
And don't forget that both accidents happened at fairly low height (not much time to fix things), when the engine would have been on a high power setting (so making things go wrong very fast), and the pilot workload isn't exactly "sat around twiddling thumbs".
Not the only vehicle. Some started with a single pot, but when the problems were realised, two pots working oppositely (i.e. 1 goes 0 to 5V from closed to full throttle, the other does 5v to 0V) became the norm.
That (as you say) allows the management system to detect an abnormality and apply some safety limits.
And it's not at all uncommon to find that zero testing took place.
One job I went to was to test the network cabling in a couple of offices that had been used by the same tenant but were now to be let separately. The outgoing tenant had got someone in to pull the cabling for one of the rooms back into that room and re-terminate it into a new panel in a new rack.
It took me a while to figure out exactly what they'd done. Every point tested half-OK - two pairs were connected, two weren't, and the two connected pairs weren't always wired 1-1. Then it twigged - they'd split the pairs because the panel had each socket connected to 4 terminals top, 4 terminals bottom, and they'd wired it 8 wires top for one cable, 8 wires bottom for the next (or it could have been the other way round, sockets wired to 8 terminals top or 8 terminals bottom).
And for good measure, a few were just in the wrong place because they'd misread the almost hand written numbers that were almost visible on the cables.
Isn't that standard practice ?
If a spokesperson says "X doesn't happen" then they can be proved wrong and accused of lying. If they say they "have a policy against X" then even if X is found to happen, then the statement isn't a lie.
Also, by "having a policy" it's easy to deflect blame onto a small group of people "doing things against policy" - which again gets the company off the hook provided they've done a good enough job of making sure there's no proof to be found that senior manglement allowed it to carry on.
In truth, Microsoft should (in 2021) play safe and switch off 'fast-startup' by default if it can see there are other partitions on the system disk
But that would mean them actually accepting that there might, possibly, in some parallel universe be some oddball who doesn't think computers exist for the sole purpose of running Windoze. OK, a bit of exaggeration for effect, but basically MS don't care if someone wants to try and run a different OS - any problems involved in doing that are nothing to do with them and there's no benefit to them of taking such weird behaviour into account when deciding what their OS should do.
Logically, they could easily have given Windoze full read/write capability for a variety of "foreign" file systems. But they didn't do that for any of them for the simple reason that it suited them, and still does, to make it as hard as possible for anyone to stray from the Windoze ecosystem.
subs haven't been double hulled for a VERY long time
Maybe not, but substantial parts fwd and aft, and on top, do have a casing that's separate to and outside of the pressure hull. Pressure hulls are circular - ever noticed that subs tend to have a flat top to them ?
If you did "drive into something" then the fwd casing would take a considerable battering but hopefully that would act like a crumple zone and avoid damage to the pressure hull end dome.
That's potentially a breach of GDPR
Err, what's with the "potentially". Try as I can, I cannot think of a practical way in which anyone could be using WA for that without breaching GDPR - badly. Just the fact that "Joe Bloggs" (note, made up name, any similarity to any real/living Joe Blogs in purely coincidental) is in Anytown Infirmary would be sensitive personal information and there's no way the T&C for WA (we can hoover up whatever we want, when we want it) would make this GDPR compliant.
What's more, it's hard to see how anyone with any contacts in their address book could use WA now without breaking the law - unless they are conscientious enough to ask every single contact for their informed and freely given consent.
Pedant alert !
In that photo, the bearings on show are standard sleeve bearings. The rollers are not bearings but drive rollers. It took me a couple of reads before I realised the mistake being made.
And I very much doubt that there's a four speed gearbox, that'll be a four step electrical control for the motor. It would simply have put the motor into a low power mode for setting off (to avoid a big jolt) on the first step, gradually increasing the power with each step.
No, it's not at all like that. OK, it's only a bit like that.
If you use Gmail, or Earth, or Google search, or ... then you are using a service without paying any money for it. That service does cost money to provide.
Analogy. You employ a butler (yes, bear with me) or cleaner, or cook, or whatever domestic service. They offer you two choices - either pay them in cash, or they can work for "nothing" but will feel free to rifle through all your private paperwork and sell your information to others. Were it like that then it would be a choice - you either pay with cash, or you pay with privacy.
But all these online services don't offer that choice, and enough people use the service to make the data slurping worthwhile. So it's not so much a case of paying someone not to rob you, it's a case of paying someone for a service vs them monetising you to pay for your use of that service. So in that respect, there's no element of paying someone not to rob you - because it's a simple case of you using a service and either paying in cash or paying in kind.
Where it breaks down is where you don't use a service. I don't use Faecesborg, have no intension of doing so, yet I know (thanks Mr Schrems) that they will ILLEGALLY have a significant pile of information on me. In respect of that, and only that, you would be correct in saying that it's like paying someone not to rob you.
Sadly, it's long past the point where any of these piles of scum would be able to convince me that I could trust them to honour any sort of "I pay you, you stop stalking me when I use your services" sort of agreement. They've all demonstrated quite clearly that their only concern is to obfuscate things for their own benefit. And they've clearly demonstrated that they think the only thing they've ever done wrong was getting caught. But roll back a decade or two, and they might have succeeded in offering a "cash or kind" offering.
What your analogy is more like is where F, G etc have their claws in almost every other 3rd-party site and harvest people's data even if they don't have a F or G account.
Which is exactly what they do. All those sites with the cute "f" logo etc - most of them include code that means Faecesborg can tell what you are reading without you having given them any permission whatsoever to do so.
Yeah, but what some seem to do is "suggest" by careful wording that the price is fixed, but in the small print is the detail that the cost will go up by CPI+3.9% each year. So the customer has accepted the contract without really reading it, and is then surprised to find that the contract didn't actually say what he assumed it would say based on the deliberately misleading, but strictly speaking a lie, advert.
You see, a fixed price contract can have price rises in it. It's just that the price is static except for defined changes - vs changeable at any time they feel like it for a "not fixed price" contract.
I'm surprised that the USA does so well
Their figures are "well fudged".
They have a system rigged to make the numbers look good. If a single premises in a block can get a high speed service then the whole block is counted as having that speed available - even though they can't have it. If you search this site, you should find some articles about it.
Tests/research has proven that the official figures are about as real as ... OK, I can't think of a simile for how rubbish they are.
But that doesn't account for the adverts that want "5 years experience in ${new technology that only appeared 2 years ago}". So the only people that can pass the sift are the bullsh**ers - honest one's will get rejected for only having 2 years where 5 is required.
And it doesn't account for the adverts that want lots of experience in loads of different things - i.e. the sort of thing many of us could do after a few decades in the industry. But then you realise the pay is for a first line helldesk droid. So honest people fail the sift because they aren't experts in a gazzillion different things, experienced people don't apply because they can see through the bull**it, and only the bull**ers get through because they blatantly pad their CVs with "creative wording".
And as the bull**iters get more brazen, the hiring droids over-egg the requirements even more to compensate, and they stop even more capable and honest people from passing the sift.
There are still use cases, but the usual problem is that you have to add on a chunk of time to get from where you want to start from to where you can start from, and similarly at the other end a chunk of time to get from where you can land to where you actually want to be.
At a previous job, we had a site on the Isle of Wight, and as you describe, getting from the north west to there was either a 3 day trip by other means, or a longish one day trip by light aircraft. For us though, the biggest hassle was the hour and a half drive to get to the airport at our end. And where I work now, they run their own mini airline flying a regular shuttle service around the country (but from a much nearer airport) simply because of the time it saves.
Oh, that sounds a bit familiar - but not on that scale.
But you raise a good point - it's no good providing a long runtime off a UPS if the systems are going to overheat from lack of cooling before the batteries give up.
At my last gig we used natural air cooling - a fan to draw air in, and another to blow the hot air out. As it took very little power, that was wired into the UPS. Running a big chiller plant would be a different prospect and, as you point out, just fixing the diesel genny is likely to be a better option.
But going back another job hat, I worked for a smallish giftware manufacturer. At the time we had a fair number of power cuts due to being on the end of a long rural electricity network. Each time we had a power cut, manglement would ask about the price of a diesel genny - what's the point of having backup for the server and phones if the lights and PCs are off in the offices ?
Each time the facilities manager would dust off the old quote from his drawer - choice of big genny, runs whole site; or little genny, runs specific services but needs investment in switching - and each time they'd decide (with the lights now back on and hence no obvious problem) it was too expensive.
Seen something similar with a client a few years ago.
Small server room, one rack for servers, one for switches and cabling. One small AC system screwed on the wall. I knew the AC was suitable as before it had been installed, I had the client forward me the details so I could check it was suitable for the continuous use and very dry air - the manual was very detailed with it's performance tables for the various combinations of humidity and temperatures :-) All worked well for a couple of years - then we got a call from the client to say there was a lot of beeping from the server room, it was getting a tad warm in there.
So they propped the door open, and directed a fan to circulate air which kept it cool enough until the AC engineer could get there. The symptoms were a bit like the system had lost it's gas - there was hint of cooling, and then the unit tripped.
To be fair, the engineer did turn up in a reasonable time. But then declared that it was the wrong sort of system. When challenged he then declared that it was "the wrong sort of room" and too much heat was coming in through the ceiling - so the clients went and got a roll of insulation to put over the ceiling which, not surprisingly, made no difference.
After a few rounds of this sort fo thing, they phoned and asked for me as one of those "knows a bit about everything" people. I had a discussion with the AC engineer who was adamant that the system could never cool that room, and he didn't care that it had been working for a couple of years just fine because it can't have been.
So I rang the AC company, spoke to the service manager and (politely) expressed my opinion that the service tech was "not making sense", the service manager agreed with my opinion, and within a few hours the fault was found (stuck reversing valve) and the unit was fixed. But until then the client had been in a standoff with the AC tech who refused to accept there was anything wrong with the system. So just getting the tech on-site isn't necessarily the end of your troubles.
Well if we're swapping overheating car stories ... the icon illustrates what an overheated car can look like.
A while ago, a friend used to do off-road rallying, and at the time I drove a Discovery (an early 3 door V8 - the "pass anything but a petrol station" engine option, except this was converted to gas to make it vaguely affordable to feed). Normally I had no problem, but unknown to me the viscous coupling for the fan had failed. So off we went, having picked up the rally motor from his mates farm, and headed ... up Shap. Well actually, we overheated before we got to the bottom of Shap - the higher load and lower road speed towing another vehicle showed up the cooling problem.
Pulled over, diagnosed the problem, let it cool a bit and topped up the water - now what to do about the fan as we faced the climb up "that hill". Rummaged around, and found a piece of baler twine in the hedge. Wrapped the loose ends round the hub of pulley and put the resulting loop over a fan blade - then started up. Rotation just took up any slack and it gripped really well, driving the fan at full speed - noticeably noisier, but kept us cool for a few hundred miles up to Scotland, around the various stages for a couple of days, and back home again. Always keep a bit of baler twine lying around, it has many uses :-)
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
