Team, Wanted to know anyone using starrocks exclusively for querying external tables? Delta/Iceberg. Want to know the performance. We are potentially looking at starrocks as our next MPP engine

It's possible to go shared data and only cache data locally - SSDs help a lot. Most folks end up using shared data config to allow for matviews on critical tables and more features. Read and write is fast on my external iceberg table setup, but not as fast as starrrocks local tables which are ludicrously fast. It's not hard to bring up Polaris and Starrocks using docker. When you're ready for the next step please keep speaking up and you'll get good answers. Don't forget to ask @Rocky whatever you want to know, he's built to know this stuff and he's got good insight. I find https://deepwiki.com/StarRocks/starrocks very useful too.

I agree with @Tony Wasson that query performance on Iceberg is not bad but, assuming you need analytic rollups, the recommended pattern is Async MV with query rewrite enabled on top of Iceberg. If you set MV to "loose" consistency you should always hit the MV instead of the base table for applied dimensions.