Re: [eman] WG Last Call for draft-ietf-eman-energy-monitoring-mib-08 and draft-ietf-eman-energy-aware-mib-13

I agree with the chairs and each energy object must belong to a single domain.

The MIB modules that have been presented and reviewed many times over,  have also stressed the importance of 1-1 mapping between  an Energy Object and a domain.

Keywords can be used as tags for other groupings of energy objects.

Thanks
Mouli

I agree with the chairs and our approach that the device/chassis is a scalar domain not a vector. As agreed in the framework model.

The complexity of power supplies (stack power for example) is handled by grouping (keywords) not the management domain( i.e. community). MUST is the preferred approach in practice.

Our implementations and experience have found that no site ever used the domain for this purpose but used groupings as Brad expressed below.

Also the information model has been expressed as scalar not a vector for this field for some time and we have feedback and consensus on running code that this is preferable.

A change to SHOULD would mean we'd have to add attributes for size of field /table to do a proper vector. That would mean soliciting feedback from implementors that already are expecting / using this as a scalar. most are not paying attention since wglc.

Change the framework error not the modelling at this point please.

Jp

Sent from my iPad
(expect ridiculous spelling mistakes)

On Jan 30, 2014, at 9:42 AM, "Brad Schoening" <brads@coraid.com<mailto:brads@coraid.com>> wrote:

Juergen,
In your dual power supply example, I would think of these as three
domains: supplyA, supplyB and supplyAB.  The new domain supplyAB is a
high-reliabliy domain.  You don't mention what the power policy of this
new combined domain is, but there could be several choices for supplyAB.
Consider a more complicated example, you have a chassis with three power
supply modules.  There are at least three different policies that could be
applied:
- Power module redundancy - total power allowed is one less than the
number of supply modules
  E.g., total power is 2 x.
- Power module redundancy with throttling - total power allowed is equal
to available power, but
  requires consumers to throttle down if one module fails.  E.g., total
power is between 2x and 3x.
- Basic power w/o redundancy - total power allowed is equal to all
available power with no redundancy.  E.g., total power is 3x.
Combining power domains likely always invokes a policy and thus in the
eman model, creates a new domain.
This approach is compatible with the chairs suggestion of MUST.
Regards,
Brad
On 1/29/14 3:41 PM, "Juergen Quittek" <Quittek@neclab.eu<mailto:Quittek@neclab.eu>> wrote:
1. Having a device limited to a single domain membership is a
limitation/feature of Cisco's EnergyWise for which in all the long
discussions we had in past year never any technical reason was given
(although it was asked for multiple times). In the contrary, I gave use
cases where it is REQUIRED to have two or more domains. An obvious
example is a highly reliable server that has dual power supply. Its two
power supply lines SHOULD be in different domains in order to be better
protected from local power supply failures. This server should be given
means to report via the MIB that it is member of two domains.
_______________________________________________
eman mailing list
eman@ietf.org<mailto:eman@ietf.org>
https://www.ietf.org/mailman/listinfo/eman
_______________________________________________
eman mailing list
eman@ietf.org<mailto:eman@ietf.org>
https://www.ietf.org/mailman/listinfo/eman