I'm excited to discover Longhorn. It seems like the answer to complexity I've encountered with solving cluster storage. Looking forward to storage tiering improvements in Harvester, and have tiering working currently with guest clusters. Looking forward, what are the limitations of Longhorn for scale? Are there well known thresholds or boundaries I should be aware of so I can prepare to adapt topology or links?

Maybe there are no precise boundaries been documented before. But I think there is a document to explain the best practices when trying to deploy Longhorn in production. And you can also refer to here to fine-tune the system settings depending on the cluster scale