I'm not sure what a "spaff" is
but great headline (though it could do with more underground virgin penetration).
GiffGaff, the O2-owned mobile network that collapsed for eight hours last week, will give £10,000 to a customer-selected charity to say sorry, and promises free calls next time its service goes titsup. The outage started at 10am on Friday when leaking water took out the operator's authentication and billing server aka the …
" ... will give £10,000 to a customer-selected charity to say sorry ... "
And would that charity donation be tax deductible? Or rather will that be claiming Income Tax relief against it?
Whilst the receiving charity will benefit either way which is always good, it erks somewhat to see a company claim generosity out of what can often be redirection of money they never had in the first place.
They paid one person more than that in their "payback" one year, payment for recruiting members or helping on the forums.
As for charities there's too many who gobble up 40 or so percent of the money in wage bills and advertising (those free pens aren't cheap) and often do things that should be state provided (which wouldn't require advertising then).
Whist I agree there are many charities who offer a service that should be state provided, I can vouch first hand for Macmillan Cancer Support offering a /fantastic/ service. Their advice and support in dealing with someone's wishes that state care was trying to run roughshod over has been invaluable.
"Building a redundant SDP would be very expensive, especially considering how rarely it would be used"
Er. Yeah. That's how redundant systems always pan out. That's the point of them. You spend money on them, equivalent to how much you spent on your normal system, in the hope you NEVER have to use them. The decision shouldn't be one of "expense" (because then you won't end up buying redundancy, backups, etc.), but surely one of service continuity. How much has this event cost you - including lost customers, free calls and the donation - and are you seriously suggesting that a redundant server in place (even if underpowered) would have cost more? If there's a repeat, ever, for any reason, it's going to cost you again and again and again every time it happens.
Which is probably why, prompted by customer anger, that post was updated with "We are not ruling out a full DR site".
If people wanted a mobile telephony system that was unreliable, they'd probably be on some other service or not bothering with mobile telephony's expense at all. And we're talking an authentication and billing server here. Surely that's *not* the sort of thing you should be leaving to chance if you want your company to thrive?
GiffGaff is increasingly becoming amateur hour. Trying to achieve something grand, but doing so by basically cutting out simple IT basics.
Actually I think that their determined approach here is mind-numbingly sensible.
The network itself works without the thing. The only thing that goes pear-shaped when it's down is the ability to bill the customers. It's cheaper to just give the calls away than it is to maintain a hot standby.
Everyone wins. They don't have to build and maintain a hot spare and their customers get free calls if it does go titsup.com. The alternative would have been finding the money somewhere, presumably in their customers' pockets.
Next:
- RAID controller fails requiring complete restoration. 8+hrs outage depending on disk size, quality and location of backups and accessibility to the machine.
- Computer goes boom - 10+ hrs outage to locate, remove, provide spare, restore data.
- Datacentre has a fire - Xhrs outage due to the fire, potential equipment restoration costs.
These are all not-unheard of things in ANY datacentre. In fact, that's all that some teams would do 24/7 for their managed servers. That's the point, they are *gambling* that this won't happen often enough to be a problem and/or cost them less than putting in a redundant system (which isn't exactly non-standard procedure) - and they are forgoing redundancy on that gamble rather than taking a calculated cost. Also, the fact that they *don't* have hardware for hot-spares suggests they don't really have cold-spares either, which further suggests that actually their backups aren't really tested all that well either (or they DO already have all that hardware and just aren't using it where it would be useful and it's sitting in a cupboard somewhere).
Today we're talking about a single incident, that's cost them 8+hrs outage + £10,000 + a lot of unhappy customers. Another outage makes it even more expensive for them, and the more outages they have, the more expensive each one gets (customers run for the hills, and they'll probably have to pay for redundant servers anyway).
The network works AT A CONSTANT LOSS without the server. That much is true. And it being free to customers in those intervals and customers using it (now KNOWINGLY so) means that'll be the most expensive part of their business if it happens again. Again, it's a gamble. I'm not sure I would want to be a customer of a company that gambles that much with its profitability. The "everyone wins" mantra excludes the costs that GiffGaff would have to pay anyway. All they are saying is that the phone call costs of a single period of downtime where customers didn't know the calls were free was cheaper to ignore than to keep a hot-spare handy. That's *not* going to be true next time there's a problem.
The alternative is to run a business like a business and cost out appropriately, especially if such downtime could (and would have, if anything else had gone wrong in the meantime) get longer and longer until it reaches the point they are issuing compensation.
As a one-off trade-off for this particular incident, it's fine. But it's highlighted a key flaw in their business expenses - and next time they may not be so lucky.
Maybe they've done the calculations, and not buying redundancy is the more sensible and cost effective decision? Yes, backups are good, and data centres explode, but there's no point in doing the calculations in the first place if you're just going to buy redundancy anyway!
Also, given GiffGaff don't own the majority of the network (O2/Telefonica do), they're not the ones that are in charge of most of the redundancy, so judging them on that makes no sense whatsoever.
> However, being on £10 goodybag, my sums suggest that GG owe me about 35p.
i bought a couple of goodybags when I first got onto GiffGaff. I soon reaslied that my usage pattern doesn't warrant buying them.
So ignoring the £15 I wasted there (my own fault), I've spent £20 on my phone bill for the nearly-a-year I've been using it. And I've got £11 credit left.
So GiffGaff clearly owe me tuppence. The thieving barstewards.
Vic.
This sort of drivel makes me laugh " This was due to a major water leak at one of our third party suppliers which knocked out both our primary and backup systems"
This is from the link on El Reg to the GG website.
If it were a back up system, it would be in a different location. That is known as 'geographical redundancy'. If it is in the same place, then it is for load sharing and load balancing.
In telecoms there is no such thing as a 'back up server', only redundancy. Unless of course it is the tape back up, but then that is only uised for service resoration when things go tits up again.
According to some of the critics, giffgaff and its suppliers don't have enough brains to design a data centre properly.
Why are water and electricity so close to each other, some even ask, as if it's so rare for buildings to have both at once.
How about the racks are not on the ground floor or in the basement, like you knowalls are guessing, but intentionally a bit higher up, but then some workmen on other plant on the roof accidentally puncture the water supplies or the reservoirs for the emergency sprinkler systems adjacent to where they come down through the roof, or something like that? Quick, bring back that Dutch lad with his finger in the dyke.
A lot of houses have a water tank in the roof space, and the owners assume that the roof keeps the rain out, so the house isn't going to get wet. But if you insulate the ceiling perfectly, so the roof space is cold, you might also have to make sure the water doesn't freeze and burst the pipes. You could heat it gently. What if you have a power cut in the middle of winter? I was reading that one data centre had something quite like this while it was mostly shut down in the Christmas holidays.
The thing is to have the backups, and restore from there. Giffgaff did so.