back to article Phased out: IT architect plugs hole in clean-freak admin's wiring design

Welcome again to On-Call, El Reg's weekly column that offers readers the chance to vent about their co-workers' ineptitude. This week, meet "Oscar", who tells us about the time he had to clean up after an overly tidy sysadmin. At the time, he was working as a consultant and IT architect for a government customer that was …

Silver badge
Pint

"NEVER assume the architect knows best!"

Understatement of the century.

40
0
Silver badge

Re: "NEVER assume the architect knows best!"

Havent we got enough different plans for house designs drawn up yet?

You'd think work would be getting scarce.

They seem to have perpetuated the myth they have to be consulted on every build.

My building inspector told me I had to get an "Architect" to do lots of complicated maths to work out if my garage roof trusses were strong enough.

Its not like I was the first person to do this. Its hardly a unique building.

I said to the inspector:

"If I had bought a sectional garage like that guy across the road would i need these calculations?"

"No" , he said , "the vendor has already had the calculations done"

"good" says me , "my trusses will be identical to those trusses , ergo my calculations have been done"

"I suppose so" says the building inspector.

Saved a few quid there!

32
0
Silver badge

Re: "NEVER assume the architect knows best!"

Havent we got enough different plans for house designs drawn up yet?

To be fair, it isn't always the architect.

Yes, we had a few back-and-forth sessions with the architect who couldn't understand our specific instructions - for example, exactly which one of our children do you think is going to agree to sleep in the smallest bedroom when we asked for the children's rooms to be identical?

Our main problem was actually the planners.

The first planner (when we were still at the "here's a rough sketch I drew myself" stage) was categorical that we could not add a half an upper floor on to our bungalow.

The second planner (who actually came to look at the site) agreed that we could, but that it would have to be oriented thusly, so we had an architect draw up a plan to suit.

The third planner thought that layout looked daft and we should orient the upstairs portion at 90 degrees to the way the second planner insisted upon. This was the design which went through several iterations with the architect and...

...the plans which were finally passed were essentially properly dimensioned versions of the sketch I'd drawn right at the start!

Grrr...

M.

36
0
Mushroom

Architect Smartitect

I find new build installs are a;;ways a disaster in the making and every time it could have been avoided if the ICT team were allowed to speak directly to the contractors.

Amongst the issues I've experienced were:

A computer suite which overheated as the aircon outlet fed into a dummy clock tower. the builders decide to put plywood panels behind the openings to prevent pigeons nesting there making the tower almost airtight and preventing the heat exchanges working.

The system ran for 48 hours before the heat buildup caused the environmental systems to power down the suite.

A new library with public access computers and staff network on every floor where the inter floor ducting was made by embedding 6 inch waste pipes into the concrete floor. this meant that there was no space for the 100+ cables needing routing thorough the floors and we had to go to fibre, not a bad move but a very expensive option at the time and one which wasn't budgeted for,

Finally getting on site to find the comms room was actually in the roof void and had sloping ceilings, we ended up with 6 different comms cabinets each of a different size and all crammed with equipment rather the 6 full height cabinets.

A university where the comns 'cabinets' were actually doorways into a service void which ran the whole height of the building making installing equipment far more perilous than expected.

Finally every single new build I have been involved in leaked. For some reason leaks always end up directly above racks of very expensive IT equipment, nylon sheeting is not part of my supply list for any new build.

We even had an issues where network cables were laid in a hurry as the building was handed over late, cables were pulled thorough quickly, connected up to network points and tested, PC's installed over night and the room handed over the following morning as staff training was starting By 07:30 all the network points failed ad t turned put the cables had been under tension and had 'shrunk' overnight.

Another item in my new build kit is now a 24 port switch and long Ethernet cable as at the network always seems to fail in at least one room when live use starts.

26
0
Anonymous Coward

"Surprised the sockets on different phases were close enough together to connect the servers like that. I thought there were rules about spacing. A server installation I visited once had the cabinets on different phases far enough apart so you could not touch 2 at the same time."

Well color your self surprised. Back in the day I worked (did I say work I meant slave labor as a summer intern) at a data center that used suns and powerpc servers . Each had a redundant power supply . Each power supply was plugged into a different phase . The thinking was if one phase had a bad day t hey other would pick up the slack and the redundant power supply would not be affected . But hey this same place that it was a good idea to have a unpaid 16 year old summer intern do admin duties of a jr tech because it was free labor. But hey for me on the plus side I learned how to properly cuss from and old cranky Scottish guy.

11
0
Silver badge
Meh

Re: "NEVER assume the architect knows best!"

"My building inspector told me I had to get an "Architect" to do lots of complicated maths to work out if my garage roof trusses were strong enough."

That is not an Architects job... that is the job of an engineer. I would not step foot in a building that had it's engineering specs drawn up by an architect.

41
0
Silver badge

That is not an Architects job

dis is true I was generalising ..

The architect just draws his fantasies on paper , like ,say , a toddler .

Then an engineer actually does all the calculations , and working out how thick / strong things have to be to span the ridiculous distances the architect dreamed up

Guess who gets the credit?

[/bitterEngineeringGraduate]

30
0
Silver badge
Headmaster

Re: That is not an Architects job

and btw , it's SET FOOT

:)

9
0
Anonymous Coward

Re: "NEVER assume the architect knows best!"

Idiot CTO: Haven't we got enough computer programs yet? I don't know why we have to hire all of these expensive programmers and software architects and testers. People make inventory management software all the time. Why when I worked across the street at $BIGCO we had an inventory management that looks just like the one you want to design here, just hire the offshore code monkeys to build it. It looks the same, ergo, the design has already been done.

10
0
Anonymous Coward

Come to think about it, properly learning how to cuss is valuable asset in IT. Swear that guy could cuss in in 20 languages .

6
0
Silver badge
Coat

desinners who needs em

You wouldnt write your own word processor would you? any more than a cricketer would make his own bat!

You'd have thought we were done with clothes designers by now too.

I mean - how many way can you make a pair of jeans?

and food! we've done food , theres more than enough recipes!

10
0
Silver badge

Re: "NEVER assume the architect knows best!"

Cough - Berlin airport - cough

6
0
Silver badge

Quite

Yesterday's NY Times carries an obituary of the engineer Robert Silman, who managed the repair work that kept and keeps Frank Lloyd Wright's Fallingwater house from falling into the water.

20
0
Silver badge

Surprised the sockets on different phases were close enough together to connect the servers like that. I thought there were rules about spacing. A server installation I visited once had the cabinets on different phases far enough apart so you could not touch 2 at the same time.

So you've not worked on larger data centre kit that requires a 3 phase feed then ?

Blade chassis, disk arrays, high end server platforms etc.

There is absolutely no issue with having the phases close to each other. The problem is about appropriate training for the people working in the data centre. A suitably motivated person who want to go would have no problems finding a couple of IEC extension cables and two screwdrivers to jam into the sockets if they really wanted to go that way.

8
1

Re: That is not an Architects job

That's because Architects are nothing more than failed artists. I once had one insist I removed the main structural centre leg in a supermarket on night shift. Because it was "in the way of his vision of the wavy roof tiling that was going in" I told him no, not happening. After three nights of this I was called in to explain why I was "not carrying out instructions according to my contract". I told them to get this ass of an architect on site and I would show him why. He actually went white...

17
1

Re: That is not an Architects job

Architect draws fantasies

Engineer does math and changes the fantasies

Engineering Technician gets the blame for all the inherited problems and finally gets it all to work two weeks later.

9
0
Silver badge

There was the architect I heard of who moved a satellite dish on the plans because it didn't "look right" there. They were ignorant of he fact that the new location had no line of sight to the satellite in question. Another was an interior designer who neglected to put more than two power sockets in a room designed to be a home office. The two sockets were located by the door and fairly useless as a result.

9
0
Silver badge

NEVER assume the architect knows best!

We techies don't. However, manglement does and they are the ones who get to talk to the architect. It's our job to clean up other people's messes and smile about it.

3
0
Silver badge

Re: "NEVER assume the architect knows best!"

Our main problem was actually the planners.

I live in a listed building and I feel your pain. I also had to provide support to a relative who was trying to replace a conservatory on their listed home in the country. The house on all sides is surrounded by a large number of trees. There is no line of site from the house to any of the neighbouring properties. I did a plan of the entire property marking in blue the immediate area around the house. However the planners rejected this saying they wanted a plan of the entire property.marked in red. The scale was also apparently wrong, mine was too large which I didn't have a clue about. When I said neither of these requirements were listed anywhere they sent through the guidance sheet (that they should have sent before) which listed them..

3
0
Silver badge

Two outlets only by the door?

That's not even code in most countries. US National Electrical Code requires no more than 6' (1.8m) between receptacles in most residential rooms.

2
0
Silver badge

re: spacing of phases in the UK

If i recall, it pre-dates AC even. If you had both 'outers' of a 200/400v to 250/500v DC system (centre earth) you had to have the meters and any accessories 6' apart so arm to arm contact was impossible.

Usually a floor per 'line'.

Something about the supply regs and anything over 250v nominal ? or was that for light sockets... old books

2
0
Silver badge

Re: Architect Smartitect

Nothing like phases being wrong etc, or ethernet wiring... but one well known contractor who Love(ll) to build social housing, at least on a development i was presented with, had handily wired all the telephone wiring on one side of the building (one socket in each room) back to a main socket in the hallway... which then had a single cat 5 (not proper CW1308 but apparently that's normal now) back to the eternal NTE, and the other half back to the kitchen socket, and a second cat5 back to the NTE. (which has ONE set of wiring terminals)

Result... 8 cables jammed in one normal phone jack (Krone IDC terminals rated for 'up to 2 connections' provided the wiring is smaller gauge) and 2 cables at the NTE with terminals rated for ONE wire only.

The openreach chappe told me they'd done similar at blocks of apartments but because it was easier, he had 7 cables to contend with at each NTE.

Some 8B connectors and a huge chocblock behind the hallway socket got the ONE working... but yeesh

1
0

It's a great safety feature where it make sense - why make it a possible failure mode when there's distance and load to accommodate doing it that way. But whether it's something like a simple high density data centre it becomes unworkable, and then literally impossible when you require diverse power feeds to even a single piece of kit for resilience.

1
0
Silver badge

I requested a minium of 4 access points, one in each corner of the room

And you admit to that in here ?

Unless you disable 2.4GHz on all but 3 of them then you are artificially causing congestion - and that's assuming no other APs nearby. On 5.8G you'd be alright provided they all pick/have configured different channels.

It's a myth perpetuated by people who don't understand the basics of wireless comms that adding more APs (especially in a small space) will "improve" the WiFi.

1
1
Silver badge

It's a myth perpetuated by people who don't understand the basics of wireless comms that adding more APs (especially in a small space) will "improve" the WiFi.

Depends on the size of the room and the number of users expected. In a typical school hall in the UK I doubt that more than two APs are needed to cover the area from an RF point of view, either at 2.4 or 5GHz and in much of the world other than the US, four channels can usually be used adjacently at 2.4 (1, 5, 9, 13) without problem. Two or more APs on different channels could - theoretically, and assuming an even distribution of users between them - improve connections if there are lots of users.

That said, I've met consumer and "prosumer" kit recently that does not do DFS and TPC which means that it is limited to four or five channels at 5GHz, so the situation isn't necessarily much better than at 2.4GHz!

M.

0
0
Silver badge

I deployed some new pdus recently(as in 0U rackmount pdus not large scale datacenter pdus). They are pretty neat as they alternate the phase on every outlet (and the outlets are color coded). pretty convenient, though the locations on the outlets could use some improvement, assuming related to the extra hardware to do the alternate outlet thing, 36 outlets on the 208v 30a 3phase though probably a good 2 and a half feet of no outlets on the bottom part of the pdu.

0
0
Silver badge

The worst I've seen was in the atrium of the "Hub" building at University. 4 floors high, with offices all up one side, fully wired for Wifi.

Including a few phone hotspots, a Wifi scanner found 262 unique access point/SSID combinations (each access point provided access to three or four SSIDs, so that's 65-85 different visible access points providing the same networks, all on 2.4Ghz)

0
0

Huh?

Disclaimer: I have zero experience with three-phase UPS...

But wouldn't there be a circuit breaker for each phase output? I know that there is one on every one of the small single-phase UPSes I use.

7
0
Bronze badge

Re: Huh?

Only if you'd asked the sparky to put them in the build.

Big 3-phase boxes I've worked with just have the output feed. Local power regulations generally decide what has to happen next. (with a splash of management penny pinching).

I've specced breakers per phase per rack with resulting impressively large power panels at the end of each row to allow for future maintenance.

7
0
Silver badge
Pint

Re: Huh?

FB asked, "...a circuit breaker for each phase..."

Yes, but typically all mechanicalky linked together. If any phase is overloaded, that CB will trip itself and the mechanical link will turn off the other two phases at the same time.

Typically.

12
0
Silver badge

Re: Huh?

"Disclaimer: I have zero experience with three-phase UPS...

But wouldn't there be a circuit breaker for each phase output? I know that there is one on every one of the small single-phase UPSes I use."

In the US each phase is suppose to have it's on panel

2
2
Silver badge

Re: Huh?

I've worked with three phase systems for a {cough} few years. I will always fit mechanically linked breakers to anything that can draw a significant current. There are some 'interesting' runaway situations that can occur otherwise - mostly based on "but nobody would do that", such as: "Well the phases are nicely balanced so there's negligible total neutral current so we can run a thinner (cheaper) line".

I've seen that, and the result when a printing press contactor lost one phase - smoke coming from a long line of ducting, and lots of (expensive) single phase kit going pop.

5
0
Silver badge
Alert

Mechanically linked breakers..

In a previous life, I managed a restaurant and also did maintenance on some restaurant equipment, including conveyor ovens. We had a satellite operation in an auto plant's commissary about 50 miles away that I also oversaw.

One evening I get a page (yes, that long ago) and the message is "Oven is down in Plant 2" So I bundle up my tools and drive 50 miles to troubleshoot the problem. After some head scratching I discover that one leg of the 3-phase power is not present. So I go hunt down one of the plant electricians, which takes quite a while, then we go on a snipe hunt for which electrical panel actually services the grimy receptacle. Eventually a panel is found and we verify that the breaker actually controls power to the outlet, but it's not tripped. It also appears to only be a 2-phase with the 2 breakers mechanically linked. The electrician verified this by taking the cover off the panel--only one screw was holding it and it came crashing to the floor when it was removed.

Now we're both scratching our heads and with the help of another electrician and a ladder he begins tracing the wiring back from the outlet. He finds that 2 phases and the neutral are wired to the panel we found, but one diverges in a ceiling electrical box and runs to a panel about 50 yards away in another part of the labyrinthine plant. There, another 2-phase breaker is found with only one phase connected, and it was not just tripped, but defective to the point where you could practically make it trip by breathing on it.

I can't imagine the level of incompetence/indifference/rushing to complete a job that would lead to wiring something this way, but it was one of the first displays of this sort that I'd experienced, contributing to barely anything surprising me these days.

20
0
Silver badge
Thumb Up

Re: Huh?

A friend works on TV transmitters. They have a UPS which holds things up until the diesel generator comes online. TV transmitters draw a LOT of current. Therefore, it's a BIG UPS. And the diesel is big, too. Especially if they serve four or five transmitters at the same site.

But, no matter how big the diesel is, if it doesn't start up, the TV goes off when the UPS uses up its batteries. Which is exactly what brought my friend out to the transmitter site that night, to discover that the people who had installed and "tested" the diesel, had not hooked it up to the UPS, so it never sensed the power failure, and never got the signal to start, even though it had worked perfectly when they tested it (with the manual start switch)

Moral: test the WHOLE system, not just your part.

12
0

Re: Huh?

What? Your standard 120v panel has 2 "phases" in it, and if industrial kit had to have separate panels for each phase, the result would be frigging huge. Beyond that, even at the utility level we have combined panels for three phase power. Why the hell you'd ever want separate panels for each phase is beyond me, it would make the protection systems basically impossible to build easily, the disconnects dangerous as sin, and load balancing and monitoring basically impossible.

I'm going to guess you've only ever worked on residential style power systems. I'd personally place input and output circuit breakers unless the damage from shutting down would be worse than setting it on fire. Now, 3 phase circuits are supposed to have their breakers ganged(linked with a small pin through the breaker toggles). It is possible to run a three phase system with over a hundred amps of current imbalance between phases, but it takes life off the equipment(and the wimpy stuff in your normal three phase UPS would kind of explode) so the utility has imbalance monitoring systems on at least the transmission lines(the big towers/cables) but the relays to control the breakers for that are $5,000 apiece.

But I understand how you could get the idea from residential power.

6
0
Silver badge

Re: Huh?

Moral: test the WHOLE system, not just your part.

Well, obviously. But how often do you get approval from manglement to risk the system going down by running those tests? If you're really, really lucky the entire team ends up working on the morning of Christmas day.

9
0

Re: Huh?

"But wouldn't there be a circuit breaker for each phase output? I know that there is one on every one of the small single-phase UPSes I use."

Transistors blow to protect the fuses, fuses blow to protect the breakers, breakers only blow as a last resort to "hopefully" stop the building from burning..

14
0
Silver badge
Thumb Up

Re: Mechanically linked breakers..

+1 for "snipe hunt"

4
0

Re: Huh?

Not always.

I've done -48V DC power systems where the rectifiers are distributed across the 3 phases and disconnected the linkage between the incomer breakers them at the client request. The reasoning was that if a rectifier failed in a big way then it would only take out one phase, not all three.

These power systems were for emergency service equipment with duplicated everything and no RCD/GFI in sight. Mind you the earthing systems were impressive.

Andy

1
0
Silver badge

Re: Huh?

If you're really, really lucky the entire team ends up working on the morning of Christmas day.

At a TV transmitter location? not likely - can't miss the Queen's Speech!

An early tuesday morning maybe, say 02:00 would seem more likely, some time when the least viewers possible could be affected.

0
0

Don't think this through

The company l worked for at the time moved in a new building with a dedicated server room. The server room had a mixture of ups fed power and normal power sockets.

When we experienced out first power cut, due to someone digging up the power cable outside the building l went down to the server room to shut down the servers in an orderly fashion. As we had been informed it would take sometime for the electricity to be restored.

Small problem although all the servers were on a ups power feed none of the monitors or kvm's were. The clock was ticking as the ups battery only enough capacity 2 hours.

Trying to figure out where to find a ups power feed whilst only the emergency lighting was on was great fun. Diving behind racks and pedastals tracing back all the spaghetti.

27
0

Ah... the bypass switch

Yes. It's a very satisfying and impressive switch to throw.

For years, I have looked for an excuse to install something equally impressive (possibly with arc extinguishers?) at home, but not having much luck...

11
0
Silver badge
Trollface

Re: Ah... the bypass switch

Take up ham radio as a hobby, get an outdoors long-wire aerial and you have the perfect excuse to connect it through one of those lovely bakelite-handle knife switches - the old radio handbook I read as a kid swore that these used to be switched over to a grounding lead in stormy weather...

6
0
Silver badge

Re: Ah... the bypass switch

That wasn't the same radio handbook that told you how to make your own meter probes (for use on valve/tube equipment no less) out of solid brass rod with a 'tape cover to protect against shocks' ?

3
0
Silver badge

Re: Ah... the bypass switch

Not sure about that, but it definitely was the handbook illustrating how to make a "detector" out of a Gillette blade and a graphite pencil lead...

6
0
Silver badge

Re: Ah... the bypass switch

Not the same book then, but it looks as if it's in similar vein

0
0

Never trust the back of an envelope

While working for a large carrier in Paris, I met with the electricians to ask as to their progress in wiring out our MDF with the requisite PDU diversity. I was assured that all the commando plugs were on the same phase and that I could power up the two Cisco 12000 GSR's that I had racked previously. I asked for some evidence of this, and was presented with an envelope on which the wiring diagram was scribbled!

Being suspicious of this, I asked for someone to connect a multimeter across two sockets to verify they were on the same phase. Funnily enough, the electricians were rather surprised to see 400V of potential difference, and I congratulated myself at having saved the company around half a million dollars, eg the cost of one populated GSR. One million if I had been dumb enough to connect both at the same time.

Great story for the "what's the most you've ever saved your previous employer" interview question :-)

52
0
Silver badge
Thumb Up

Re: Never trust the back of an envelope

Trust...but verify!

// ALWAYS verify.

19
0

Re: Never trust the back of an envelope

By a quick finger in the air calculation... probably somewhere in the region of about £10M. Twice.

2 separate incidents where I was moonlighting as a weekend operative (while employed as a Product Engineer during the week) which ironically made me the highest ranking person in the building both times. Both times we had a sudden spike in failures of transistors at test. Cost of silicon, about 20p each, cost of final device about £20. Both times resulted in tracing the fault back to the silicon enabling us to test and dispose of the silicon instead of the finished product.

In typical arrogant fashion we were still forced to use one batch of silicon as the US wafer fab refused to update their test limits, despite the obvious savings to the company as a whole, but as our factory was UK based we were treated as a customer and not part of the same company.

This was the same company who then moved manufacturing to Mexico and demanded all the test equipment be supplied without safety cut-offs, which could have resulted in the Mexican staff being electrocuted. Needless to say the manufacturer's response was, "no safety cut-off, no machines"

25
0
Silver badge

Re: Never trust the back of an envelope

>Trust...but verify!

Never Trust....

2
0

Page:

POST COMMENT House rules

Not a member of The Register? Create a new account here.

  • Enter your comment

  • Add an icon

Anonymous cowards cannot choose their icon

Forums

Biting the hand that feeds IT © 1998–2018