BOFH: Tripping the mangelfreuzer switch
Crash, bang, wallop
Episode 25 BOFH 2004
"..and so the point of this meeting is to decide on the service level for your equipment, so we know what you're expecting when we purchase your new box."
"Well, 24 by 7 obviously," one of the assembled beancounters responds.
It never, EVER, amazes me at how a box in the throes of dying sharpens the minds of the beancounters - particularly when it's their box that's going down every half hour or so.
"Ordinarily I'd agree with you, but seeing as the company would appear to be 'beancounter rich', with none of you starting before 9 or staying after 4, I'm wondering if this is the most cost effective way of providing coverage for your machines?"
"We do run batch jobs out of hours," another beancounter adds drily.
"Yes, a weekly pay preparation for the HR machine, a weekly report, a monthly projection plus the annualised runs - from what I can determine. The annual combined total of which is about 40 hours runtime..."
"So you're suggesting what exactly?"
"I'm suggesting that as we're about to replace your financials box you might want to revise your service level to determine if you really want to be paying for 24x7 maintenance when you could probably get away with 9x5 or less?"
"Couldn't you people just maintain the equipment and reduce the cost further?"
"We could, but there's always the chance the mangelfreuzer switch will trip - which would be a problem."
"What's a manglef.. r."
"Manglefreuzer - as in Mangle-Froy-TSir. It's a switch or other device that hardware manufacturers put into their equipment to detect unuathorised access to a box - and makes the system act strangely."
"Yes, you know, the odd memory error, harddisk write error, machine crashes, etc. To encourage you not to tamper with the box. And generate revenue for the maintenance provider of course."
"Are you serious?!"
"Of course. Some of our major outages have been those exact same problems."
"You mean they're all caused by a manglefreuzer?"
"Some of them, yes. Obviously no sane vendor would build in something which could be conclusively linked as the source of a problem, so they make the outages sporadic and hard to track down..."
"And people know about this?"
"Technical people, yes."
"And you're ok with that?"
"We don't really have that much choice in the matter, do we? Anyway all the vendors do it - and not just in computers."
"I bet that's what's wrong with my home machine!" one beancounter adds, accompanied by the nods of other would be system builders.
"Yes," I comment. "It's bound to be that and not your lack of antistatic protection, earthing and poor installation method. Those vendor bastards!"
I avoid the topic of the PFY's habit of onselling dodgy RAM to company staff as this could muddy the discussion somewhat...
"So what you're saying is you can't look after the kit properly and you're recommending that we buy a machine that has a warranty that matches our real need for server availability?" one the beancounters asks - basically reiterating the content of the email message which asked for this meeting.
"Yes," I sigh.
"And does this come at a premium?"
"Any extra service comes at a premium, but when you take into account the potential impact of not doing it - like the outages you've been having - it's a small price to pay."
"So what are we talking about?"
"It's about triple the price for a five-year onsite warranty."
"TRIPLE?!" a beancounter gasps.
"BUT, if you reduce the service level to 5x9, 4 hour response, it's only about twice the price."
"TWICE! Is that what we've paid in the past?!!!"
"No, we've paid nothing in the past, which is why you've had to spend so much time waiting for engineers to show up for machine failures."
"Yes, I suppose we realise that now - but surely machines should last longer than a year?!"
"Entry level desktops, you'd get four years out of - I guess. But leading edge hardware like yours is pretty special and as computer vendors say - the candle that burns twice as bright burns half as long..."
"I... well how much is it going to cost?"
"All up, maybe 4000 quid."
"Oh! Well that's not as bad as..."
"Plus maintenance," I add, carpe dieming.
"Worst case scenario. I mean we might be able to get the vendor to drop a couple of grand if we pushed it - but no promises..."
"How much do you think?"
"Maybe - if we can really get them over a barrel - 6k - but like I say, no promises."
"Really?" the Head beancounter gasps, semi-happily.
"Well, I suppose we can go with that, and the level of support. But how long would it take to get a stable system back again?"
"Delivery time on this kit - if we're lucky - could be something like two weeks if they have the units on hand"
"And how long to change over to the new machine?"
"If we pulled an all-nighter we could probably do it in one - if it's urgent."
. . .
Later, in Mission Control.
"So how'd it go?" the PFY asks.
"Two new top-of-the-range desktops and a stack of overtime!!!"
"Just remember to put a note in your diary for a fortnight away to remind you to stop powercycling their server every time you go into the computer room."
"And start mentioning the manglefreuzer problems on the Human Resources server - I fancy a new laptop."
"Good as done..." ®
Sponsored: Benefits from the lessons learned in HPC