get real (oper) power consumption of switch via SNMP

[fhk-cwempe]

I want to read the power consumption of our switches (e.g. C4510).

currently I am using these OIDs:
System Power: 1.3.6.1.4.1.9.9.117.1.1.1.1.4.16
Inline Power: 1.3.6.1.4.1.9.9.117.1.1.4.1.4.16

But at least the Inline Power does not seem to be the real power consumption.

A-04-S1#sh power inline
Available:3272(w)  Used:954(w)  Remaining:2318(w)

Interface Admin  Oper            Power(Watts)     Device              Class
                            From PS    To Device
--------- ------ ---------- ---------- ---------- ------------------- -----

Gi1/1     auto   on         10.2       9.5        IP Phone 8961       4

A-04-S1#sh power inline police
Available:3272(w)  Used:954(w)  Remaining:2318(w)

Interface Admin  Oper       Admin      Oper       Cutoff Oper
          State  State      Police     Police     Power  Power
--------- ------ ---------- ---------- ---------- ------ -----
Gi1/1     auto   on         none       n/a        n/a    5.0

[...]

--------- ------ ---------- ---------- ---------- ------ -----
Totals:                                                  450.2

The OID mentioned above returns 954 Watts.

But the actual power consumed by my PoE devices is 450 Watts according to "show power inline police".

When I dimm the display of a Cisco phone I can see "Oper Power" decreases for that port.

So I think these "Oper Power" values are the correct ones.

We want to implement some power saving features and I would like to monitor the changes.

But how can I read the real power consumed by the switch (total or just inline power)?

[fhk-cwempe]

I just noticed I do get real values from some of my switches. (e.g. C2960XR-)

Using this OID: 1.3.6.1.2.1.105.1.3.1.1.4.1

But using this OID on a bigger Switch (C4510), I only get the values per module.

.1.3.6.1.2.1.105.1.3.1.1.4.1 = Gauge32: 128
.1.3.6.1.2.1.105.1.3.1.1.4.2 = Gauge32: 164
.1.3.6.1.2.1.105.1.3.1.1.4.3 = Gauge32: 16
.1.3.6.1.2.1.105.1.3.1.1.4.4 = Gauge32: 167
.1.3.6.1.2.1.105.1.3.1.1.4.7 = Gauge32: 94

But even theses values seem to be a little off (+30%).

And I would have to sum up these values to a total number, which is not easy with MRTG.