[BLUG] NOV meeting topic

[David Ernst]

Dare I say it?

25 power plants to double the number of computers in the world seems
like a bargain to me.  Looking at:

http://www.industcards.com/ppworld.htm

I roughly counted how many power plants are in the world right now,
and easily got over 2500.  So, a 1% increase in world power plants
would enable a 100% increase in computers..?  That's quite a return on
investment, and that's using the numbers that seem high by Mark's
reckoning (which makes sense to me).  

Looking at it another way, apparently Indiana and Ohio together
produce roughly enough electricity to power all of the worlds
computers.  Probably a little bit short, but... these two states are a
sliver of the world's land area, and the air here is still clean even
though we're burning a lot of dirty coal in old plants.

Suppose we wanted to double the number of automobiles in the world.
We'd likely be looking at something like an 80% increase in power
consumption.  That's a rough estimate based on zero research, but
whatever the correct number is, it's going to be way way way over 1%.
And a car takes WAY more energy when it's in use!  

http://auto.howstuffworks.com/horsepower2.htm

Says that a ford escort uses 110 horsepower = 82,026 watts.  So,
driving a Ford Escort (my old one used to get about 33 mpg (need I
point out that the escort is not among the highest performance
vehicles ever designed?)) for ONE MINUTE is roughly the same power
consumption of leaving a desktop computer idle (at 60 watts) for ONE
DAY.  The twelve hours of driving I'm planning on this weekend to
visit my parents will the amount of energy of two years of leaving my
desktop turned on 24/7.  And cars are mobile things with difficult
emissions control problems.  The computers' new power plants could be
anything from wind farms to fuel cells.

If we want to conserve energy, we're much better off focusing on
vehicles than computers.  Meanwhile, we're probably all drawing more
power for lighting than we are for computing (60W?!?  I've got
individual light bulbs that draw that much! (though fewer and fewer of
them)).

Having said all of that, I love the idea of thin clients, and I see it
as a minor tragedy that they aren't more popular.  But, I would focus
on deploying them in business environments.  Gone would be the days of
people wondering whether to save to their computer or to the
server... and if their computer broke, just give 'em a new one, have
them log in, and they're right back in business...  Save power?  All
the better!

David



On Thu, Oct 11, 2007 at 11:10:44PM -0400, "Simón A. Ruiz" wrote:
>Wow, lots of thoughts. Awesome.
>
>ben lipkowitz wrote:
>> A typical desktop will draw about 60W when idle, and a typical laptop
>> will draw 25W at idle. Let's face it, most computers are idle most of
>> the time. Both types of computers draw near zero when in sleep or
>> standby mode. Many computers are not configured by default to use sleep
>> mode, and people are not going to figure out how to enable it. Then
>> there is the short delay in starting up which leads many people to turn
>> off this feature. A computer could "learn" its user's habits so that it
>> will already be started up by the time they wanted to use it. So, this
>> is a software problem really.
>
>I disagree. There is definitely a software variable in this equation,
>but do we need to be drawing 60W for an idle computer, when we could be
>drawing 5W for an active computer? No amount of software hacking will
>make that change.
>
>In the move to alternative energy sources, there are many variables in
>the equation. Software is one. Hardware is one. Power supply efficiency
>is one. The efficiency of solar cells, wind turbines, hydro-electric
>dams in creating the energy. The inefficiency of transmitting that
>energy long distances and the need to decentralize power generation.
>
>Human habit is the biggest.
>
>We need to examine every single Watt in and every single Watt out,
>because spread across 5 billion humans, every Watt matters tremendously.
>
>The world will not step into the light (to borrow from a famous science
>fiction writer) until it is not only economically feasible to do so, but
>economically stupid not to.
>
>Let's ignore the environmental impact of every single Watt we use. What
>about getting computing up and working in places off the grid? Places
>where you might have a solar panel or a guy running a bicycle (it's
>happening) to generate all the power for miles around. Do you use a
>computer that draws 60W when it isn't doing anything? Or twelve
>computers that draw 60W collectively when twelve people are using them
>to learn, to connect, to communicate, to drive business, to create
>art...to get lost in the Wikipedia?
>
>Chris Colvard wrote:
>> I was thinking about this awhile ago.  Now think of these thin-client
>> front ends being provided by an ISP with the ISP providing online
>> storage space and the typical apps people use (Word Processor,
>> Spreadsheet, etc.) as AJAX applications (or something else hosted).  If
>> the ISP owns the thin-client then a subscriber doesn't have to manage
>> software or buy new computers and the ISP probably gets a much easier
>> support environment since the thin-client's only function is connecting
>> to the ISP's servers.  Thoughts?
>
>One thought: http://www.koolu.com/
>
>Incidentally, less than 10 Watts, running full force.
>
>Mark Krenz wrote:
>>   That's not right.  Sorry Simon, nothing personal, but he's off on
>> those figures.  He's using big numbers to wow people and I think that's
>> a bad thing to do.  He should use real figures.
>
>Ack! It would lose the power of simplicity, get boring, and still make
>the same point (only to a less engaged audience) if he tried to make a
>scientifically accurate estimate up on stage. It was a numbers game, not
>an objective statistic, and I felt that was made clear.
>
>We could go through the intellectual exercise of coming up with a more
>accurate figure, but you have to admit, we'd be off on that as well.
>
>I'm sure the figure we're tossing around here of 60W for an idle
>computer could be analyzed further, but that would detract from the
>basic point of the conversation, which is not an intellectual exercise
>in computer power usage; at least it isn't for me.
>
>The basic point of this conversation, at least insofar as what I said
>regarding computer power consumption is that computers as we know them
>today are inefficient and impractical for use outside of a first-world
>infrastructure, as we look towards inviting previously unrepresented
>folks into the global digital world.
>
>Which also happened to be the basic point of maddog's talk.
>
>>   Still, I know what he's getting at and conservation
>> is a good thing so its good to think about.
>
>Well, what he's getting at is that it's more than just "a good thing",
>it's unavoidable as we look towards the future of technology.
>
>In my opinion, beyond the future of technology, it's a matter of
>survival for me and my family, you all.
>
>>  Sorry for being down on him
>> but I would have thought Maddog would be more careful with his
>> statistics than that.  I mean, people quote him.
>
>Yeah, you're probably right, though I'd be surprised if anyone quotes
>his number game as an authoritative fact.
>
>It was certainly useful in bring this list to life. ;-)
>
>>   Now let's say I have a Xen server that has enough RAM and CPU power to
>> host 64 machines (16 GB of RAM and 8 cores would do nicely).  This
>> machine has a 700 Watt power supply.  With all the virtual machines
>> running it would use a maximum of 700 watts, maybe even less.  There
>> would be a little more latency overall, but the average watt usage per
>> virtual machine is only 11 watts.
>
>Precisely my point.
>
>However, drop the Xen server. Add an LTSP server, and you can probably
>server somewhere in the 200 or more thin-clients at less than 5 watts
>per, add less than 10 watts for a Koolu thin-client front end, and
>you've got less than 15 watts per workstation. (Leaving out the monitor,
>of course...)
>
>That's a dramatic increase in efficiency. And my bet is that the end
>users won't notice the fact that they don't have a full workstation's
>processor-time at their command. Firefox and OpenOffice would come up
>like lightning, since they've already been loaded into memory by one of
>the other 200 people sharing that terminal server.
>
>How long would it take for that setup to pay for itself in power
>savings? A couple of years?
>
>Say 5 years later, you decide you want to upgrade the system to handle
>Ubuntu Pretty Parrot's amazing new on-by-default holographic desktop
>environment (humor me), What does it cost to upgrade all 200 of those
>workstations? The price of a new server.
>
>And the migration can be done with no disruption, just point the DHCP
>server at the new thin-client server, and have everyone turn off their
>thin-client at the end of the day.
>
>I think workstations like what we have will only be used by geeks like
>us in the not too distant future. People who don't just use computers as
>a portal to the Internet are the only people who wouldn't be 100%
>satisfied with a thin/thick-client solution.
>
>>   Doable and done.  Matt Liggett, a former sysadmin at Kiva and the guy
>> who wrote the original knowledge base at IU, was working on a program
>> for IU that would slave the idle time of desktops into one giant super
>> computer.  I'm not sure how far he got with it, but I've heard of other
>> places doing this before.
>
>Is that the OpenMosix project?
>
>As I understood it, it only worked with specific software. You couldn't,
>say, offload any old video encoding or (place random mundane
>high-processor-time task here) job yet.
>
>But, yeah, talking to that guy at the IU LinuxFest the year before last
>got me thinking about merging a cluster of physical computers into one
>processor pool and basically make each of the computers a thin-client to
>that collective "machine".
>
>I think that this type of thing has the potential to become big.
>
>Steven Black wrote:
>> That sounds like the definition of "Internet Appliance".
>
>*SnIp!* Nod. Well, yeah, what do most people use a computer for?
>
>> In an environment where people expect Desktop-like computers, it is
>> likely to be a hard sell. However, if you do what Símon was talking
>> about, and go overseas and sell it to an entire building at a time...
>> It may be doable, especially if you're dealing with an environment
>> in which most people are not expected to have computers at all.
>
>Again, http://www.koolu.com
>
>Check it out.
>
>Joe Auty wrote:
>> Perhaps all of this will come to fruition regardless with Google Docs
>> and Spreadsheets?
>
>No, seriously ;-) check out http://www.koolu.com that's their whole shtick.
>
>It's a $200 no-moving-parts box, draws less than 10W of power, that can
>run full-fledged Ubuntu, except that in order to keep the processor load
>low, they suggest you use Google Apps on Firefox rather than load both
>Firefox and OpenOffice at the same time. This is also a solution to the
>problem of not wanting to have a hard disk locally. You CAN buy a $300
>version with a hard drive, but that removes a few of the advantages of
>getting a no-moving parts, low-wattage box.
>
>They also work really well as thin-clients, in which case (with a decent
>back-end), they'd function almost indistinguishably, on less than 10W,
>from a computer that idles at 60W...completely indistinguishably when
>it's idle. ;-)
>
>Their main markets are abroad, of course, because we Americans like
>driving Hummers.
>
>Anyhow, yeah, I'd like one or two to play with.
>
>Thanks for all the responses, I appreciate this dialog.
>
>Have a good night!
>
>Simón
>_______________________________________________
>BLUG mailing list
>BLUG_at_linuxfan.com
>http://mailman.cs.indiana.edu/mailman/listinfo/blug

--