__/ [ JEDIDIAH ] on Thursday 05 October 2006 15:14 \__
> On 2006-10-05, ed <ed@xxxxxxxxxxx> wrote:
>> On Thu, 5 Oct 2006 07:13:47 -0500
>> JEDIDIAH <jedi@xxxxxxxxxxxxx> wrote:
>>
>>> > One uses traditional load balancing setups, Linux-HA, IPvs, CARP,
>>> > round
>>>
>>> That still doesn't get around the boundary condition of
>>> everything peaking at once or things peaking past their projected
>>> maximum. Excess capacity and "margins" exist in engineered systems
>>> to work around the problems of not fully understanding your problem
>>> or perhaps how to solve it.
>>
>> One builds some contingency into the network, just like we did pre-vm,
>> one cannot have too high risk for a single component. VM's allow one to
>> spread backups around.
>>
>> It's just like thinking of a computer inside another computer... we just
>> put more than one Russian doll inside each. Doesn't mean we have less
>> Russian dolls.
>
> Sure it does.
>
> Server consolidation is the perfect excuse for management to skimp
> on capacity. They can dellude themselves into thinking they have enough
> extra because the consolidation process will make them think they have
> more than they reall do by skewing statistics.
>
>>
>>> Computing simply isn't mature enough as a discipline
>>> (engineering or otherwise) to operate on tight tolerances.
>>>
>>> > robin DNS, NAT round robin, there are loads more less traditional
>>> > means also.
>>> >
>>
>> Computing is a very mature subject.
>
> Get back to us in 6000 years.
>
> Computing isn't even out of the womb yet.
It's all relative. One could argue that the hindrance Moore's Law is facing
is enough indication that we're hitting a brick wall. The greater problem at
the moment is innovating, software-wise. There's lack thereof, surely. There
are many splendid things we could have implemented, but they are not
necessarily what makes the most /money/, so there's suppression and
resistance to advancement. And no, Google is far from breaking that barrier.
There's complacency too. All they do is acquire some companies that never
reinvented a wheel. And face recognition is nothing new, either.
Best wishes,
Roy
--
Roy S. Schestowitz | "Computers are useless. They only solve problems"
http://Schestowitz.com | SuSE GNU/Linux ¦ PGP-Key: 0x74572E8E
run-level 5 Sep 28 23:16 last=S
http://iuron.com - help build a non-profit search engine
|
|
