Any detail reason on this?
Yes, a database is the hardest thing to scale horizontally. Zeebe uses stream-processing and event-sourcing as paradigms, rather than mutation a centralised database.
See this talk by Deepthi from CamundaCon 2019: https://www.youtube.com/watch?v=JbXKgQQmukE
And this article by Bernd: https://blog.bernd-ruecker.com/zeebe-io-a-horizontally-scalable-distributed-workflow-engine-45788a90d549
Thanks，this is helpful.
I think , maybe Zeebe could use TIDB which is an opensource distributed HTAP database as storage in the future. https://github.com/pingcap/tidb