4 ms·
This could be done by mounting S3 as filesystem and plain sqlite. Any differences?
by Datagenerator 4y ago
This could be done by mounting S3 as filesystem and plain sqlite. Any differences?
- simonw 4y agoIf you did that, wouldn't you have to fetch the entire SQLite database file before you could perform any queries? With this you don't need to do that - the software magically fetches just the pages needed to answer your specific query.
- spullara 4y agoThat really depends on the S3 filesystem implementation. It could make the same sort of optimizations as this does using range reads.
- michalc 4y agoAh even if this were the case, in some situations you can’t mount filesystems (e.g. Docker without certain capabilities enabled?), and so something more in the application layer would be the only way to access SQLite on S3?
- samatman 4y agoYep! This kind of thing works around these problems. https://www.sqlite.org/draft/useovernet.html https://www.sqlite.org/draft/useovernet.html