Re: [FRIAM] Cloud computing (russell standish)

2009-10-18 Thread Pietro Terna
   This is my first intevention in this list; I enjoy very much reading 
it (I was at the Santa Fe Complex in July, hi Stephen ...).


   My point of view is that what we need is also a good grid services: 
I've now a model which requires two hours to run on a standard quite 
fast pc; for each experiment I need to run it upon 18 different sets of 
parameters and our local voluntary grid ( 
http://ramses.di.unipmn.it:8080/status/WebStatusServlet ) solves very 
well the problem.


   Best, Pietro Terna


--
The world is full of interesting problems to be solved!
Home page http://web.econ.unito.it/terna



FRIAM Applied Complexity Group listserv
Meets Fridays 9a-11:30 at cafe at St. John's College
lectures, archives, unsubscribe, maps at http://www.friam.org


Re: [FRIAM] Cloud computing (russell standish)

2009-10-18 Thread Owen Densmore
Hi Pietro!  Nice to see you on our list.  We met in Italy at U. of  
Bologna a few years back.


Welcome!

-- Owen


On Oct 18, 2009, at 1:19 PM, Pietro Terna wrote:

  This is my first intevention in this list; I enjoy very much  
reading it (I was at the Santa Fe Complex in July, hi Stephen ...).


  My point of view is that what we need is also a good grid  
services: I've now a model which requires two hours to run on a  
standard quite fast pc; for each experiment I need to run it upon 18  
different sets of parameters and our local voluntary grid ( http://ramses.di.unipmn.it:8080/status/WebStatusServlet 
 ) solves very well the problem.


  Best, Pietro Terna


--
The world is full of interesting problems to be solved!
Home page http://web.econ.unito.it/terna



FRIAM Applied Complexity Group listserv
Meets Fridays 9a-11:30 at cafe at St. John's College
lectures, archives, unsubscribe, maps at http://www.friam.org




FRIAM Applied Complexity Group listserv
Meets Fridays 9a-11:30 at cafe at St. John's College
lectures, archives, unsubscribe, maps at http://www.friam.org


Re: [FRIAM] Cloud computing

2009-10-18 Thread Owen Densmore

On Oct 17, 2009, at 4:51 PM, russell standish wrote:

The whole cloud thing as presented in this article leaves me a bit
cold.


Few of the articles are written by folks using cloud computing, thus  
its just buzzword compliant hype.


There really are some interesting cloud things happening, but I  
haven't seen a good article yet on it.  After using Amazon, its  
clearly just hardware in the sky and not all that exciting.  Mainly  
useful for moving your IT infrastructure outside of your buildings.


But GAE (Google App Engine) has serious interest.  First of all, it  
has a much easier way to build web apps.  It does NOT give you LAMP  
-- but it does give you an interesting replacement which makes  
building web apps much simpler.  And further, it integrates all the  
Google Desk Top apps and other services and libraries (Maps, Earth,  
Analytics and so on) into the GAE engine.  The Google File System with  
Big Tables is pretty wonderful too.  World wide secure transactions  
with a fragmented, replicated, file system.


A third interesting cloud environment is Aptana, more programmer  
oriented and comes with an Eclipse plugin/front end.


Unfortunately, very few articles capture the cloud scene.


I don't want software as a service, I want it as an application,
running on my own computer with my own data.


Well, clearly cloud computing is not oriented toward desktop/personal  
computer applications.  So sure, you wouldn't want cloud computing for  
your situation.



With open source, I can
get the applications at the price I can afford, and adapt them if
needed for my needs.


Agreed .. but again, the target is not you.  Its folks who want to  
shoot their computers and servers and migrate away from the gawd awful  
problems of maintaining their own IT infrastructure.


Don't forget: one of the biggest problems companies is how noob their  
employees are.  The discipline of cloud computing is compelling for  
companies not wanting their spreadsheets walking around in laptops  
that travel world wide.



If I need serious grunt, then no cloud will solve
the supercomputing problem - regular high performance computing
centres are still needed for that (although if the Grid is ever
delivered not still-born, that is an alternative).


The best we've got so far in this area is cloud (web) render farms,  
very popular and inexpensive for Blender users.



I can see some point in enterprise-wide clouds though...


Yup.  Actually, if I were building a small business today, I'd go  
cloud, via the for-pay Google system.  I'd fire anyone having a  
company spreadsheet or document on their laptop and not in the cloud.   
Or company email on their computer (POP rather than IMAP).  It is just  
too expensive and dangerous.  I was chief scientist for Sun's IT  
department for a couple of years and the things civilians do would  
blow your mind!


   -- Owen



FRIAM Applied Complexity Group listserv
Meets Fridays 9a-11:30 at cafe at St. John's College
lectures, archives, unsubscribe, maps at http://www.friam.org