On 08/07/2018 07:28 PM, Patrick Donnelly wrote:
> On Tue, Aug 7, 2018 at 4:29 AM, Joao Eduardo Luis <joao@xxxxxxx> wrote:
>> On 08/07/2018 11:18 AM, John Spray wrote:
>>> On Mon, Aug 6, 2018 at 8:05 PM Sage Weil <sweil@xxxxxxxxxx> wrote:
>>>>
>>>> On Mon, 6 Aug 2018, Yehuda Sadeh-Weinraub wrote:
>>>>> Option #3?
>>>>>
>>>>> "3 objects"
>>>>> "23k objects"
>>>>> "3.4M objects"
>>>>> "2.3k PGs"
>>>>
>>>> Yeah I like that better. It's slightly inconstent with the EIC units, tho
>>>>
>>>> "1 B"
>>>> "3.2 KiB"
>>>> "2.3 PiB"
>>>
>>> I like option 3 too. I think it's okay to have a different format for
>>> counts ("1k objects") than we would for distances/times ("1 km")
>>
>> Agreed. Especially because one specifies counts (e.g., 1k objects) and
>> the other specifies units (e.g., 2.3 PiB).
>>
>> And anything else kinda feels weird.
>
> Technically speaking, "objects" is a unit. (Compare to Mole as an SI
> unit which is defined count of atoms/molecules.)
That's a fair assessment, but I'd argue we'd need an agreed upon
quantity per object to consider the "object" as a unit, much like mole
is bound to the avogadro constant.
However, an object is a blob of data, which in our case can be of any
size depending on who's writing it. So we do have a representation of
space being consumed in the cluster when we talk about an object, we
just have no idea how much.
I'd say an object is to Ceph the same way a tupperware container is for
soup. We could make "tupperware container" a unit, but that will give us
nothing of true value regarding the amount of soup.
-Joao
--
To unsubscribe from this list: send the line "unsubscribe ceph-devel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
