[irtf-discuss] Re: [Green] Re: [E-impact] Network Equipment Energy Efficiency Metric

[Tony Li <[e-mail]>]

[-green, as this is inappropriate]


> On Nov 12, 2024, at 8:25 AM, Toerless Eckert - tte at cs.fau.de <[e-mail]> wrote:
> 
> I was primarily thnking of provisioning powr for peak use. Or as necessary do
> peak shaving. For example, if there atually is a significant peak after e.g.
> a power failure, as there is in traditional servers with spinning disks, then
> this can typically be shaved through software/control mechanisms. I am not aware
> of anything like this for traditional network devices, but always good to
> know upfront that there is no such situation where power use is higher than
> power use during e.g.: 100% utilization, for which one does plan the power supply
> of the room/


Conventional high-end routing equipment already does a staged power-on sequence in order to reduce surge current at power-on. This is isolated to the system itself and does not span across systems. At cold boot, fans are necessarily spun up to full power to avoid worst-case situations. Once the ambient temperature can be ascertained and enough software is running to ensure that the system is not in thermal distress, fan power is reduced to maintenance levels.

Data sheet power usage for networking gear is typically peak, worst-case power utilization, as this is necessary for provisioning purposes.

T