Le 2012-02-02 05:26, Invisible a écrit :
>>> I would have thought that needing more disk space is such a crushingly
>>> rare event that it makes almost no sense to optimise for it.
>>
>> No it's not. You may need to run an app in debug mode for a while and
>> need extra space to store the dumps. You may run into seasonal peeks and
>> need extra storage just for that period. Etc... Sizing hundreds of
>> servers for a worst case scenario is not efficient use of the company's
>> money. You are much better off having some amount of slack space that
>> you can swing around when needed.
>
> Maybe that's the problem. I cannot imagine any task that would ever
> require "hundreds of servers". I'm struggling to think of a task that
> would require more than about a dozen at worst. I mean, unless you work
> for one of the largest companies on the planet, which almost nobody does.
>

While it's true that most tasks performed by the Fortune 1000 rarely 
take more than a dozen or so servers.  Each of those companies have 
dozens of such tasks.

For example:
- A bank of web servers for their corporate web site
- A handfull of DNSes
- A couple entreprise directory servers, authentication servers, etc...
- 1 mail server per 1000 employee
- 1 file server per 1000 employee
- A dozen servers for the HR apps
- A dozen servers for the Finance/accounting dept.
- A couple dedicated server for the law dept. (Confidentiality)
- Server monitoring servers.
- Network monitoring servers.
- Environment monitoring servers (Heating/air conditioning, power 
consumption, generators, battery banks, etc...)
- and then... more servers for the actual business that company does, be 
it putting people on planes, managing people's money, building cars, or 
selling drugs.

Split those across a few data centres for geographical proximity to the 
user base and duplicate all the mission-critical servers.

Tada!

I don't understand why you somehow think one needs to actually work for 
British Airways, Siemens, Yamaha or IBM to be able to imagine what's 
it's like to work in a company with 100,000+ employees.

>>> You claimed that it's not insane to run a SAN over the Internet,
>>
>> When did I say that?
>
> Francois Labreque wrote:
>  > We've also had the performance discussion before. Yes, the
>  > theoretical access speed of a local SATA drive is much faster than
>  > that of a SAN attached logical disk, but in actual real world
>  > practice, with real world data, there's not much of a difference,
>  > even on SANs located halfway across town in another building.
>

"In another building" DOES NOT mean "over the Internet".  Private data 
circuits predate the internet, and no company would use the Internet to 
link two data centres together.  Remote offices, maybe, but definitely 
not data centres.  15 years ago, it would have been done via data-grade 
telco circuits (such as an E1 or E3 line, usually called a WAN circuit). 
  Nowadays, it's done via DWDM optical circuits (usually called a MAN or 
Metronet).

>> Not that SANs are infaillible, but why do you assume that it will fail?
>
> They want to back up 200GB of data per day over a 5 Mbit/sec Internet
> connection. You do the math.

So their backup strategy will fail.  Not the actual SAN.  Gotcha.

>
> Plus, since when does rolling out a brand new complex system go
> smoothly? ;-)
>

When it's planned properly.  I've seen it happen.  But, I must admit, 
that I've seen quite a few disasters, too.

>>> So what you're saying is that a handful of the richest companies on
>>> Earth can afford to do this?
>>
>> There's more than a handful of companies who can afford it. A few £M in
>> extra telco costs per year is nothing compared to the prospect of going
>> out of business because your data centre had a 110-story building crash
>> on top of it.
>
> So you're saying that more than "a handful" of telco companies exist?
>

There are, in fact, more than a handful of telcos in most countries. 
while the actual wires underground usually belong to one or two 
companies, the other ones will buy bandwidth from the 
cable-owning-entity and resell it to other corporations.  This is why, 
for example, I can rent circuits from BT in Canada, even though they do 
not own any of the fibre optics here.

But this has nothing to do with the fact that I was talking about 
businesses buying telecommunications services from the telcos, not the 
telcos themselves.  Last time I checked HSBC was not a telco, and I do 
know that they've had 10Gbps links between their data centres for over 
10 years.

-- 
/*Francois Labreque*/#local a=x+y;#local b=x+a;#local c=a+b;#macro P(F//
/*    flabreque    */L)polygon{5,F,F+z,L+z,L,F pigment{rgb 9}}#end union
/*        @        */{P(0,a)P(a,b)P(b,c)P(2*a,2*b)P(2*b,b+c)P(b+c,<2,3>)
/*   gmail.com     */}camera{orthographic location<6,1.25,-6>look_at a }

Post a reply to this message