Originally from the User Slack
@Alexandre_Debril: Hello Scylla people
We’re designing our data model in such a way that we suspect we will end up with very small partitions (like 1 to 10 rows / partition) spread across nodes. Do you think it can be an issue ? Is there any recommendation we should follow for that kind of data design ?
Thanks in advance
@Felipe_Cardeneti_Mendes: it primarily depends on the queries. If you frequently need to read 100 rows, you would need to run 10 queries. 1K rows, 100 queries. And so it scales and may add to your client-side tail latency
But yea, data distribution wise, this looks good
@Alexandre_Debril: ok, great. Thanks for your answer @Felipe_Cardeneti_Mendes