Project
Databricks' managed serverless Postgres database service, with autoscaling computes, storage separate from compute, and instant branching.
Databricks tests Lakebase Search on 1M to 4M rows, using a 23,000-row internal app as its baseline. A search request touches two computes, the app's and the database's, and either one can set the limit.
Reality
Databricks says Lakebase restores Postgres in seconds even at 100 TB by creating a branch at a timestamp. The figure times branch creation, so it matches a finished RDS restore only if that branch serves production reads at once.
Reality
Hot pages on large fixed-size Lakebase computes now sit in Postgres shared buffers backed by huge pages instead of falling through to the local file cache, and Databricks measures up to about 2x throughput.
Reality
A p99 of 200 milliseconds from Kafka to the online store beats batch cadence by a wide margin. It also stops before your model reads the feature.
Reality
Coatue's round values the company at roughly 27 times a self-reported $7 billion run rate, about what December's round paid. What changed is that the acceleration has stopped.
Reality