Add first draft of indexing ADR.

[tpendragon]

Refs #30

[hackartisan]

This is very clearly laid out, and I like the way the performance reasoning is in each section.

This descriptions in this document focus primarily on incremental changes (although I think the performance metrics are oriented around a full cache refresh in each case). We haven't talked as much about a full refresh of all 3 caches. Should we try to futher specify that process more here or address it separately?

[hackartisan]

Is this the same log order value as the previous table, or different?

[tpendragon]

This descriptions in this document focus primarily on incremental changes

That's interesting, it was totally my intent that this ADR be mostly about a full cache refresh, with a note that incremental changes should be much faster. I'll have to re-read with this viewpoint in mind.

[hackartisan]

This descriptions in this document focus primarily on incremental changes

That's interesting, it was totally my intent that this ADR be mostly about a full cache refresh, with a note that incremental changes should be much faster. I'll have to re-read with this viewpoint in mind.

I think it's that each description says "every minute..."

Also I think it would be good to talk about the life of the event log. if we do a full cache refresh, then another one, does the event log just get bigger? Do we know where to start to read the latest of everything?

[tpendragon]

Oh good read, we need another column that's a consistent ID so we can compact it. So that if it exists already we UPDATE.