Per-Entity Metadata Working Group

[Ian Young <>]

> On 7 Aug 2016, at 22:36, Tom Scavo 
> <>
>  wrote:
> 
> Is such a strategy advisable?

There are a couple of obvious downsides. The more minor one is the need to 
not just add a couple of aggregates, but to provide these alternatives for 
each stage of the metadata maturity pipeline. So, it's really adding two for 
the production aggregate, two for preview and two for fallback. If you don't 
do that, anyone for whom the split actually solves a problem is locked out of 
using the preview and fallback aggregates making their situation less robust 
as the trade-off for less resource usage.

The more significant downside in my view (and this has always been my 
position every one of the numerous times this has come up in the decade since 
the UKf moved into production) is that this is not a real fix for anything. 
It's not a strategy, it's a band-aid, and implementing it will allow people 
to push fixing the real problem off into the future again.

To a certain extent it's about perception, though. This is most problematic 
if the reason people want you to do it is to temporarily sidestep the 
resource requirements of large aggregates. It's much less likely to mislead 
people into thinking that problem is fixed if the headline is that it's a 
"discovery aggregate". You'd only need one new aggregate in that case, too.

    -- Ian

Attachment: smime.p7s
Description: S/MIME cryptographic signature