I'm curious if there will be any public discussion of the k-anon server data operations. Given it'll be a critical dependency of many businesses, ours included, I'd be interested in things like:

• Sharding and Replication: I could see sharding the K counts by IG owner being helpful as it could result in fewer ad techs experiencing failure in the case of server outages or other issues.
• Geo Replication and Consistency: I'd imagine values will be updated immediately in the nearest DC and then eventually replicated to other regions? Will it replicate to all regions globally, just "local regions" (like US East to US West), or would that be configurable somehow.
• Querying and Monitoring: will an ad tech have any insight into metrics relevant to their K values, numberOfObjects, numberOfUpdates, numKTooLow, numKAccepted etc. I understand we'd want to limit somewhat for privacy to avoid interesting attacks, but I'm also envisioning debugging without some of the metrics we're used to and can see that being helpful. Rounding values or some limited delay would be fine.