Designing high throughput server is hard. One of the problems we often encounter is how to update internal data structures that sustain mostly read traffic.
Paul E McKenney conducted a thorough comparison of most performant implementations within this scheme. One of them is a cooperative algorithm called "Quiescent-State-Based Reclamation".