About the upsert performance to updata a DFS table

Question

I am now testing dolphindb, now I create a distributed table and want to update it. I wonder how about the performance of the "upsert" function to update this table?

Answer 1

Upsert is suitable for low-frequency update scenarios. Its underlying implementation uses the Copy-on-update mechanism, that is, it will first copy the partition data on the relevant column, and then perform the modification and replacement operation, so the time consumption will generally be at the scale of seconds. If there are many partitions involved, it will take more time.