Use protocol 2 with extreme caution, it is not yet production ready.
It does variable length chunking for inline deduplication, with sparse
indexing on the server side.
It can deduplicate the same blocks in differently named files across
multiple clients, whilst being memory efficient due to the sparse indexing.
For more details
on the ideas and development, you may read the report
that I wrote for my degree in PDF or
online HTML format.
Note: the report refers to 'protocol 1 / protocol 2' as 'burp 1 / burp 2'
It is currently missing some features, such as Windows EFS, encryption and
The status monitor and counters are also not fully working.
Do not expect protocol 2, and the backups that it makes, to be compatible
with future releases of burp.