- GemStone images can reach terabyte sizes, much larger than the largest comfortable 7GB Pharo image.
- GemStone uses an object table to map object IDs to data pages on disk for efficient loading of objects into virtual memory as needed.
- Transactions log modifications to disk before committing to ensure data integrity, while concurrent transactions are allowed if they don't modify the same objects.
- Identity-based and indexed collections provide fast access to large collections without needing to fault in all elements.
- Developers can use Pharo for initial development and then port applications to the more robust GemStone/S deployment environment.