[LINK] 'Cloud Computing doesn't fly'

[grove at zeta.org.au]

On Tue, 28 Sep 2010, Marghanita da Cruz wrote:

> grove at zeta.org.au wrote:
> <snip>
>> I wonder if a couple of my 2008 predictions are sort of coming to pass?
>> 
>>> 6) Supplies of rare earths, metals and noble elements will become 
>>> extremely scarce.  Stock prices for these will skyrocket, tech drivers
>>> will be put on hold and commodity tech goods will inflate or become 
>>> impossible to get.   An IT equipment demand will emerge and vendors will
>>> be forced to use what they have.  Business will innovate and extend 
>>> lifetime of existing systems until the new tech becomes affordable again.
>>> The second hand IT market rockets up and Linux finds a new market in these 
>>> areas.
>> 
>> (not quite, but I was on the right track)
>> 
>> And:
>> 
>>> 9)  Virtualisation becomes mainstream, with purpose built engines provided 
>>> with VM ready OS available as a default.  Substandard components will 
>>> cause major "grid failures" as dozens of VM nodes collapse on failing 
>>> hardware based on "6".
>> 
>> If the Virgin computer crash was caused by a virtualisation failure,
>> expect more of "9".....   Virtualisation is of course, the key to making 
>> Clouds work.....
>> 
>> Then again I may well be talking through my hat, if I had one......
>> 
> <snip>
>
> Is there any evidence that old computers become unreliable due to hardware 
> failures?

Absolutely.   The components used are prone to failure, such as capacitors 
in power supplies or on various connection boards.    I am not talking about 
commodity computers here, but "big iron".   We bought a bunch of servers 
in batches ran them flat out for 4 years and sure enough, when one failed, it's partner 
failed similarly 6 months later.  Likewise when disk drives are purchased in batches. 
We had a RAID storage cabinet that I replaced two disks in 4 weeks apart before 
I convinced management to get rid of it.  We are actually in the middle of a refresh 
right now, moving off 4-5 year old systems.

> Their demise is more usually due to new software and hardware 
> incompatibility. Ofcourse, the problem then is new software is less reliable 
> than old software. Though a Harvey Norman salesman once tried to offer me an 
> extended warranty on my video camera in case the software failed. ;=)

In the UNIX world, OS life cycles are not so relevant.  We can run the latest 
Solaris for example on a computer 5 years old, depending on the architecture,
but not 6 years old....

But, YMMV because the correllation is that we have some systems nearly 15 years old 
that I keep expecting to die, but do not.

The truth is, I believe that there is a quality problem with systems coming out of China.
In the last 3 years, I have noted that reliability of some components is going down, 
while older systems just keep working.   There are noted instances of component substitution, 
such as using lower spec or tolerance parts in place of mil-spec components and so on.

And some products we relied on for a long time are being replaced by systems that are certainly 
less stable or have early life failures etc.....

I put all this down to the fact that there is a shortage of rare earths, a booming greed 
economy leading to substitution and dumbing down of some design practices and quality control.


-- 
Rachel Polanskis                 Kingswood, Greater Western Sydney, Australia
grove at zeta.org.au                http://www.zeta.org.au/~grove/grove.html
    "The perversity of the Universe tends towards a maximum." - Finagle's Law