What is it about?
It is an extension of Parallel Snapshot Isolation (which is itself based on Snapshot Isolation) that makes different sites in a distributed database see the same transaction history, unlike other algorithms implementing PSI that are usually built under the assumptions of data partitioning. In addition, it is also implemented using only two timestamps, unlike other solutions where the number of timestamps is the same as the number of sites.
Featured Image
Photo by İsmail Enes Ayhan on Unsplash
Why is it important?
It allows systems running under PSI (whose advantages are presented in the paper) to be applied in disaggregated database architectures, which are more and more common especially in cloud computing environments (e.g. Amazon Aurora). In addition, the two-timestamp solution avoids the performance drawbacks of the multi-timestamp ones. namely in materializing the snapshot and computing transactional conflicts.
Perspectives
I think TOPSI could be an effective tool in order to make distributed database systems under strong consistency more flexible and affordable.
Nuno Faria
Read the Original
This page is a summary of: Totally-Ordered Prefix Parallel Snapshot Isolation, April 2021, ACM (Association for Computing Machinery),
DOI: 10.1145/3447865.3457966.
You can read the full text:
Contributors
The following have contributed to this page
