[nycbug-talk] Larry Ellison on "Cloud Computing"

pete pete at nomadlogic.org
Mon Sep 29 14:54:11 EDT 2008



On Mon, 29 Sep 2008 14:48:50 -0400, Siobhan Lynch <slynch2112 at gmail.com>
wrote:
> 
> On Sep 29, 2008, at 2:11 PM, pete wrote:
>>>
>>
>> well - i'm not sure about that; but i think the point is this: the  
>> basic
>> operating principle(s) are not that new, and have been tried by a  
>> lot of
>> companies over time.  i think the joking bit is a side effect of the
>> cynical marketing that surrounds these efforts.
>>
>>
>>> While I somewhat agree - amazon is still in pioneering territory -
>>> lots of companies built data rooms in their companies then came  
>>> Exodus
>>> and other colocation providers - lots of people built their own
>>> "grids" or virtual machine computing with hypervisors and such in
>>> their datacenters - now Amazon is in the same space metaphorically as
>>> the colo providers - all they need is a decent SLA and a way to
>>> recover fast.
>>>
>>
>> they my be pioneering in how they market this, or in their billing  
>> model;
>> but i remember working with IBM years ago to run my applications on  
>> their
>> managed clusters/grids/cloud etc, in IBM run DC's.  I'm sure IBM was  
>> not
>> the first to do this either.  unfortunately - for us we found that  
>> this did
>> not scale for us from a pricing/performance perspective.
>>
>> FWIW - this is coming from an HPC environment so it may have worked  
>> for
>> some other companies out there with different business models....but  
>> it
>> certainly did not work for us.
>>
>> -pete
>>
>>>
> 
> 
> See, I always thought IBM's thing was on thier own OS's and you were  
> locked into running on AIX or OS/400,, or whatever, where Amazon seems  
> to allow you to run whatever images you can get running on the EC2  
> system.
> 
nah - we could put any OS on it we wanted, and could choose from Intel or
PPC hardware.  they handled rebooting/hardware maint/RMA's etc and provided
a more realistic SLA than the Amazon offering.  the only problem was that
for what they charged it didn't make sense for us to run any real
production code, or simulations/renders on the kit they would have
provided.  maybe if we where a hedge fund or something like that with a
couple sim's running on a small cluster, and an ORA instance or two it
would have worked.

-pete

-- 
Pete Wright
pete at nomadlogic.org
310.869.9459




More information about the talk mailing list