[Beowulf] Fwd: Recycling old nodes without poisoning Indian
Robert G. Brown
rgb at phy.duke.edu
Sat Apr 16 08:26:43 PDT 2005
Robert G. Brown http://www.phy.duke.edu/~rgb/
Duke University Dept. of Physics, Box 90305
Durham, N.C. 27708-0305
Phone: 1-919-660-2567 Fax: 919-660-2525 email:rgb at phy.duke.edu
---------- Forwarded message ----------
Date: Sat, 4 Dec 2004 15:39:45 -0500
From: Bill Taylor <wataylor at as-st.com>
To: Robert G.Brown <rgb at phy.duke.edu>
Subject: Fwd: Recycling old nodes without poisoning Indian chlidren
Begin forwarded message:
> From: "John Carroll" <carroll at dsuper.net>
> Date: December 4, 2004 2:16:31 PM EST
> To: "Bill Taylor" <wataylor at as-st.com>
> Subject: Re: Recycling old nodes without poisoning Indian chlidren
> Another site that may be useful to Robert Brown:
> This provides lots of references by category of scrap.....he may wish
> to click on one of the many links such as
> Add or View listings for Computer Scrap and Dismantling
> and follow the links from there to track down and discuss with
> business and process participants.
> ----- Original Message -----
> From: Bill Taylor
> To: John Carroll
> Sent: Friday, December 03, 2004 4:29 PM
> Subject: Fwd: Recycling old nodes without poisoning Indian chlidren
> Begin forwarded message:
> From: "Robert G. Brown" <rgb at phy.duke.edu>
> Date: December 3, 2004 9:42:56 AM EST
> To: Bill Taylor <wataylor at as-st.com>
> Subject: Re: Fwd: Recycling old nodes without poisoning Indian chlidren
> On Fri, 3 Dec 2004, Bill Taylor wrote:
> I am beginning to get some information about recycling computer nodes.
> I'll keep forwarding this to you until it appears that we either find
> way or approach diminishing returns. Let me know when that appears to
> be happening.
> I'm hoping to actually spend a day revising and adding to the book
> fairly soon now, as winter break approaches, although I'm going to be
> very busy indeed for just about one more week. I'll incorporate
> whatever has built up in email like this by that time;-)
> I might eventually write a cluster world magazine column on it.
> What I REALLY need is the comparative economics of it, and as you note
> the critical thing there is shipping costs. Obviously nodes can be
> shipped "slow boat to china mode" -- as cheaply as possible -- but the
> very cheapest ways of packing cases and shipping them are likely to be
> tens of dollars each PLUS an even more expensive 20-30 minutes of time.
> What is really needed is a rented panel truck and hired driver -- ship
> 100's of nodes up just "thrown into the truck" for $1000 or so.
> $10/node plus the time required to derack a node, strip out any locally
> reuseable parts, and carry the guts out to a loading platform on a
> is what I envision as being the upper limit of the comfort level of
> cluster operations. The time is then the unavoidable minimum -- the
> same even to throw a node in a dumpster. The cost is low enough that an
> organization might absorb it or find a way of backcharging it to grants
> at original purchase time ("prepay your node recycling fee"). We do
> pretty much the latter now with monitors, at $10/each, so it is a model
> I know can be extended to nodes and other systems on campus.
> Everybody wants to do good and avoid bad, but there are always limits
> what they are willing to pay for it. I'd like to be able to present
> some sort of economic argument that indicates that the cost isn't,
> really, all that great.
More information about the Beowulf