[ih] Greener computing

[Toerless Eckert]

On Thu, Sep 01, 2022 at 01:10:21PM -0700, Dave Taht wrote:
> On Thu, Sep 1, 2022 at 12:26 PM Toerless Eckert <tte at cs.fau.de> wrote:
> >
> > Dave:
> >
> > The sad reality is that older hardware, whether router or PCs uses
> > more jule of energy per Megabit. So, in most cases it does pay off energy
> > wise to buy recent hardware.
> 
> While the total energy/megabit might have improved, in general,
> devices are eating more
> energy than they were a decade back, I used to be able to run
> wireless-n at well below 2w,
> and more recent devices are well past 8-12w, admittedly able to push
> 10x more traffic. Idle numbers, I haven't looked at in a while.

There is certainly a lot of unnecessary energy wate in newer product
lines, because the more power you need/have at peak, the more you
need to invest into design to use less energy in idle.

If it would have just been routers, i think hardware would have
never been optimized to minimize energy consumption in idle. This as all
driven by battery driven devices, phones, tables notebook. And even
if the same hardware and OS are used in routers, simple misdesigns on
the application side can easily invalidate power savings.

> > I would assume that energy saving in PC/router hardware will be something
> > users will look a lot more into in the coming years. So there may be
> > some good logic to stick to old hardware and hope an upgrade in 3 years
> > will be a lot more energy saving than one now.
> 
> To me - "energy saving" is mostly industry code for "more unfixible
> offloads from the main cpu", and an EE employment act. Vastly prefer
> simple, open source, code and a bare minimum of those crippling
> offloads, with enough (power saving) cores, much more is feasible.

Yes, unflexibility an non-software-upgrade-option is what i hate
the most too. But i have also come along a lot of features that could
only be added because of programmable offloads. espeecially for anything,
one could classify as "QoS-related". Admittedly of course most of this
not in home-routers, but at Gbps speed, including offload (e.g.: FPGA) 
code for NTP/PTP, resilient ethernet ring L2 techniques or other FRR mechanisms,
performance monitoring, TSN/DetNet PREOF, but also all the interesting AQM.

> IF we engineered stuff for a 8-15 year field life, and got off this 2
> upgrade cycle it would be a better world.

Lets see, when smart phones are being sold like old mainframe/supercomputers,
like the CDC cyber in University back in the 80th/90th, where the upgrade
to a 3x faster machine was merely a 1 million dollar tape with new
microcode firmware without NOPs ;-) 

Cheers
    Toerless

> > Btw: If you're interested to discuss more of energy saving in networking,
> > then maybe chime in on the "recipe at ietf.org" mailing list, this is an old
> > list thart became stale, but a bunch of us at ietf114 felt it was the easiest
> > available list to discuss energy in networkin related interests (maybe up to
> > a point where someone in IETF leadership thinks a better/newer list would be
> > appropriate).
> >
> > Cheers
> >     Toerless
> >
> >
> >
> >
> > On Thu, Sep 01, 2022 at 12:12:42PM -0700, Dave Taht via Internet-history wrote:
> > > On Thu, Sep 1, 2022 at 11:51 AM Olivier MJ Crépin-Leblond via
> > > Internet-history <internet-history at elists.isoc.org> wrote:
> > > >
> > > >
> > > >
> > > > On 01/09/2022 18:41, Bob Purvy wrote:
> > > > > I think by the 90s he's on a PC, but I haven't really said anywhere.
> > > > > Definitely he's gotta try those AOL coasters.
> > > >
> > > > Remember it was floppy 3 1/2 in disks first which were terrible as
> > > > coasters, only to become more useful when they switched to CDs. That
> > > > being said I really hope these ended up somehow elsewhere than in a
> > > > landfill as so many were produced - see
> > > > https://nowiknow.com/remember-all-those-aol-cds-there-were-more-than-you-think/
> > > >
> > > > I wonder how that kind of production would fare in today's
> > > > environmentally conscious world....
> > >
> > > I kind of feel that way about home routers. There are easily 10x more
> > > of those than are turned on, that could be upgraded to open firmware,
> > > enabled with ipv6, bufferbloat fixed, made more reliable and secure.
> > > Instead they languish in junk bins or landfills. I keep hoping to find
> > > a charitably suited VC (hah!) willing to back this concept:
> > >
> > > https://docs.google.com/document/d/1T21on7g1MqQZoK91epUdxLYFGdtyLRgBat0VXoC9e3I/edit
> > >
> > > With the current supply chain problems everywhere, perhaps people will
> > > relearn how to be frugal.
> > >
> > > Recently an ISP I work with reflashed 700 "obsolete" mikrotik
> > > wireless-n routers to current openwrt and is giving them away as meshy
> > > APs. These are well built, and with current software, remain useful
> > > for another decade, at least.
> > >
> > > I haven't bought a "new" computer in years. Desktops (with enough ram)
> > > got good enough for me a decade ago.
> > >
> > > > Best,
> > > >
> > > > Olivier
> > > > --
> > > > Internet-history mailing list
> > > > Internet-history at elists.isoc.org
> > > > https://elists.isoc.org/mailman/listinfo/internet-history
> > >
> > >
> > >
> > > --
> > > FQ World Domination pending: https://blog.cerowrt.org/post/state_of_fq_codel/
> > > Dave Täht CEO, TekLibre, LLC
> > > --
> > > Internet-history mailing list
> > > Internet-history at elists.isoc.org
> > > https://elists.isoc.org/mailman/listinfo/internet-history
> 
> 
> 
> -- 
> FQ World Domination pending: https://blog.cerowrt.org/post/state_of_fq_codel/
> Dave Täht CEO, TekLibre, LLC

-- 
---
tte at cs.fau.de