How does Milvus's Timestamp Oracle (TSO) ensure operation ordering?

The Timestamp Oracle, managed by RootCoord, assigns a globally increasing timestamp to every write operation as it enters the system, giving Milvus a consistent, cluster-wide notion of "before" and "after" for operations that might otherwise arrive at different nodes at slightly different real-world times.

Because every insert, delete, and search request can be associated with a specific TSO timestamp, Query Nodes can precisely determine which data was committed as of any given point in logical time, which is exactly what makes the tunable consistency levels possible: a search specifying a particular guarantee timestamp only needs to confirm it has processed data up to that timestamp before returning, rather than needing some vaguer, harder-to-reason-about notion of "recent enough."

This design mirrors a pattern used in other distributed systems (like Google's Spanner) that need a coordinated sense of time across independently-operating nodes without relying on perfectly synchronized physical clocks, which are notoriously difficult to guarantee across a real, geographically or even just logically distributed cluster.