diff --git a/Trino-Iceberg-Integration.md b/Trino-Iceberg-Integration.md
index e586acc..308c518 100644
--- a/Trino-Iceberg-Integration.md
+++ b/Trino-Iceberg-Integration.md
@@ -35,6 +35,16 @@ s3.aws-secret-key=YOUR_SECRET_KEY
 s3.region=us-east-1
 ```
 
+### Recommended: `iceberg.unique-table-location=true`
+
+Keep Trino's default `iceberg.unique-table-location=true` (already shown in the configuration above). With it enabled, each new table is created at a unique path `s3://<bucket>/<namespace>/<tableName>-<UUID>/`. The UUID suffix gives you three operational properties worth keeping:
+
+- **Orphan-resistant `DROP` + `CREATE`** — if a `DROP TABLE` cleanup partially fails (filer error, crash mid-cleanup), a fresh `CREATE TABLE` with the same name lands at a new path and is not blocked or polluted by leftover files.
+- **Crash-safe aborted creates** — a `CREATE TABLE` that wrote partial state but failed to commit leaves debris under one UUID; the retry uses a new UUID and proceeds cleanly.
+- **Defense-in-depth against client-side empty-location pre-flight checks** — Trino runs `listFiles(<location>).hasNext()` before commit, and so do some other engines. With unique paths the check sees a fresh empty directory regardless of any historical state at the deterministic name.
+
+If you prefer the cleaner `s3://<bucket>/<namespace>/<tableName>/` layout (one directory per table, no `<tableName>-<UUID>/` sibling), setting `iceberg.unique-table-location=false` is safe against SeaweedFS — you give up the resistance properties above. Pick whichever trade-off fits your operational model.
+
 ### Multi-Level Namespaces
 
 To use nested namespaces (e.g., `db.schema`), add: