OT: cheap colo space in Southern Germany/Munich

[Eugen Leitl]

On Sat, Nov 24, 2012 at 10:27:22PM +0100, Garry Glendown wrote:
> On 24.11.2012 20:18, Eugen Leitl wrote:
> > Whether you have to fire 99% or 98% of your customers is not 
> > really going to save your business.
> Just wondering - have you already started looking for a new power
> supplier, seeing that most/all German power companies are increasing

Of course, I reconsider my energy provider choice yearly. Including
entirely other options. You don't? You should.

> prices come January 1st? What about cost of living - have you already
> started looking for a new place to live to reduce your hourly costs? ...

Why, yes, of course. You don't? It must be sure nice to not make
thoughts about that.
 
> Sorry, but customers have gotten much to used to prices for technology
> only going down. Pushed by big players like DTAG and others' dumping

In IT, prices going up are unprecedented. You might have
seen the impact post-flood Thailand, and I have a hunch
cloud customers will take a dim view upon IPv4 scarcity
driving up prices. They will rather route around that,
and I have a hunch IPv6 will be not the means for that.

> prices for Internet access, selling to end customers at prices well
> below what they are charging other ISPs. Instead of charging prices that

I would be so happy to pay cash for fiber, so I don't
have to be nickled and dimed for my pains afterwards.
But, saidly, FTTH deployments are limited to end customers,
and result in asymmetric customer-scale pricing.

> cover for both operation of current infrastructure as well as improving
> and advancing their technology, they have reduced their support and
> services to a bare minimum, ignoring necessities of the future. Like
> IPv6. After all, there is lots of time, and sooo many addresses still left.
> 
> And what do the customers do? With the oh so coveted "Geiz ist geil"
> mentality (sort of "greed is terrific"; an advertisement slogan of a
> German electronics chain) customers flock to the cheapest provider, even

I'll be looking at how this charge-customer-for-IPv4 circus
is going to work out, but only as a spectator. I have a hunch,
the show will suck.

> if they only save a Euro or two. Oh sure, things go wrong, their lines
> get disconnected, they don't get decent customer service, bills and
> account drafts may be wrong, they complain and write about the trashy
> providers. But they still stick to the cheapest place they can find. I
> bet if you go through the list of active v6-providers, most of them (at
> least in Germany I suspect) are smaller and regional providers. And

I support my regional provider. It's not my fault they can't
deliver the product I need, and I don't have the funds for 
a back up line, just because.

> quite a few of them probably already had v6-addresses on 6bone before
> official v6-addresses were distributed.
> 
> Of course end-customers aren't the ones that should be driving the new
> technology. Providers ought to, as they understand (or should) the
> implications of the technology. So we did. Having v6 in our backbone for
> well over 6 years, we ran a series of articles in our customer
> newsletters, explaining the technology and reasons (in layman terms),
> offering support in planning and implementing v6, the whole 9 yards.
> Given that we mostly provide services to business customers, most of
> which are using v6-ready equipment, one would expect it should result in
> several customers go and get v6-ready. Especially as some of our
> customers have production, suppliers and customers in the far east as
> well as Australia.
> 
> Total number of responses, even after talking to some of them on
> occasion: 0. Zero. Null. Zilch. Oh, except for one customer that asked
> us to get some v6 PI as soon as it was available. Not that they actually
> wanted to roll it out, but just to not have to renumber at some point in
> the future. But they also ordered that about 2 or 3 years ago.

I hear your pain. I very much sympathize.
 
> Of course, apart from the hen and the egg, there's also the dinosaur (of
> cause everybody except GOPs know there actually was a predecessor of the
> hen), namely the hardware suppliers. You can't blame only the customers
> or the providers without putting blame where blame is due. E.g. $C has
> been big in talking about putting out v6-capable hard- and software. To
> a point. Sure, our old 7206's already had a v6-capable IOS, and even
> their SOHO routers work fine with it. And they have for many years. But
> even such a big proponent as they is only doing at best a half- at ssed
> job. I'm not deep enough into the RFCs etc. to conclusively state how
> well v6 is implemented in their mainstream software, but apart from
> basic routers, the rest is more or less crap. Or completely missing.
> Sure, rudimentary v6-support is present, but come on - selling brand new
> datacenter switches without full v6 features (or any, for that matter).
> Other vendors aren't any better, especially in the consumer area one
> must wonder how come companies that must sell quite a decent number of
> device to the far east markets could wait so long before offering
> v6-capable devices. If they even do already. I dare to say that v6
> deployment may have looked a lot better if every router sold over the
> last 3 years had been fully v6-capable. And came with a big warning
> sheet, stating for the customer to kick his provider's behind if they
> didn't supply v6 addresses to them. Without that, big providers just
> shrugged their shoulders - no demand, no v6. Why spend money if nobody
> except geeks can run it or want it?
> 
> But I reckon I'm preaching to the choir here. Apart from maybe a select

Given this forum, definitely.

> few who'd rather fix the v4 problem not by using a working solution
> which may cost some money, nor by paying more for v4, but by suggesting
> a new solution which doesn't exist, won't work (or at least won't
> scale), and will cost money. Most likely more than the first two solutions.
> 
> As for the cost of IPv4 addresses - I dare to predict other providers
> will follow suite. Maybe not tomorrow, maybe not (yet) next year. But

This started with OVH, who started to charge ~1 EUR/IP. 
Hetzner is just engaging into price discovery, trying to figure
out how much existing customers are willing to put up with
before splitting. The decision has been made, it's just
about finding the right price.

Of course other providers will follow suit, and customers will
follow, since virtualization allows you fast-flux deployment, so
you're free to change providers at a suitable frequency.

Assuming you care, as end customers are only interested in the
cheapest possible product, and are completely technology-agnostic.

> they will. If the resource is dwindling, prices increase. That's regular
> market regulation. Supply and demand. So if you believe switching your
> provider will keep your rates down - go ahead. And be prepared to do so

I'm not switching my hoster. I'm going out of the game.

> more often. But I guess you'll get experience in migrating your hardware

Having your hardware is dead. I will retire my servers in 2-3 years,
and won't buy another ones. It's dead, Jim.

> to a new place, renumbering your servers, updating your DNS entries

Customers don't care about numbers, only names.

> (some of which may be controlled by your customers), ... by the time
> you've switched for the fourth or fifths time, you may have your
> workflows down to the point that the time spent will only equal to two
> or three times of what you save in the few months you stay with that
> provider, until he also charges for the IPs you use.

I don't see why virtualized providers can't shift their hosts in
realtime. Their only limits is contracts.
 
> In short - if a euro or two for an IP is breaking your business model,

Or, five, or ten, or hundred.

> you don't have a viable one. And if you can't pass the cost on to your
> customers, they're too cheap for decent service.

I didn't. I gave my customers notice of termination.
 
> Just my 0.023€ (prices do go up everywhere ... ;) ) Time to get off of

Try BTC.

> the soap box ...