Re: [E-impact] [OPSAWG] [IVY] side meeting #119: Power Metrics: concrete usage example

[Michael Richardson <[e-mail]>]

{not sure where this discussion should be, but the 22 CC's probably don't
need it}

Simon Leinen <simon.leinen=[e-mail]> wrote:
    > We are adding continuous and consistent power consumption monitoring
    > across our fleet of routers.  Most of them expose their overall energy
    > consumption (as well as that of some components) via the wonderful
    > ENTITY-MIB/ENTITY-SENSOR-MIB or close vendor-specific cousins of those.

    > But (even within one vendor) the representation of these sensors varies
    > widely, which makes it complicated to (a) locate the relevant/best
    > sensors and (b) deal with the different representations - some devices
    > have separate volt/ampere sensors, others have watts, some measure
    > performance into the PSU, some out of, some both.  Not to mention funny
    > bugs, e.g. claiming the power sensor is in "dBm" units where the values
    > clearly indicate Watts (280 dBm aren't really plausible :-).

Sounds like an XKCD 927 problem.
(Should I be concerned to have memorized that number?)

    > It's painful and already cost me a lot of time, which leads me to
    > suspect that router power consumption is not widely monitored in the
    > industry (I could be wrong though, maybe people use other methods).

I'd sure like to monitor it more for my limited set of equipment, if only to
see if there are unexpected spikes.
Like, do gbps L2 broadcast loops (no STP on that device) consume more power?


--
Michael Richardson <[e-mail]>, Sandelman Software Works
 -= IPv6 IoT consulting =-                      *I*LIKE*TRAINS*