Balancing eventual consistency with performance in a high-volume microservice architecture?

Hey everyone, I'm working on a system with multiple microservices that need to maintain high performance under heavy load. We're considering implementing eventual consistency for certain data flows to improve write performance and reduce latency. However, I'm a bit concerned about the operational complexities and potential impact on user experience if the eventual consistency window becomes too long. For those who have experience with this, what are your strategies for monitoring and tuning eventual consistency to ensure it doesn't negatively impact performance or data integrity too much? Any specific metrics or tools you've found useful?