Ultimately, they selected CockroachDB, as described in "RFD 110: CockroachDB for the control plane database" [1]. Now, published just yesterday, "RDD 508: Whither CockroachDB?" [2] discusses what to do in light of Cockroach going proprietary.
I'm getting increasingly nervous about baking in assumptions on software that doesn't have a strong guarantee of staying open source. Foundations like the CNCF help here, where projects are assigned ownership to the foundation.
There is a middle ground though. It's not practical to only build on top of foundation-owned software. So a metric I use these days is to consider the cost of replacement or fork maintenance, combined with how likely I think a license change would be, as part of evaluation.
Even with the Apache foundation therenis no guarantee they’ll be maintainers for support and active development. Only guarantee is a company that is profitable but doesn’t want more money…
Maybe people should be getting nervous that the whole flower power mentality of FOSS doesn't really scale when there are bills to pay, and families to nurture.
There is a middle ground though. It's not practical to only build on top of foundation-owned software. So a metric I use these days is to consider the cost of replacement or fork maintenance, combined with how likely I think a license change would be, as part of evaluation.