Improved Open Source Backup:
Incorporating inline deduplication and sparse indexing solutions
G. P. E. Keeling
4. The first iteration
The plan for the first iteration was to replace the core of the original burp
with an inline deduplicating engine that uses variable length chunks. In the
prototype, I had already produced the basic parts of this engine, so the
problem became how to merge this into burp.
Due to lack of space in the main report, the design documents that I wrote for
the first iteration are reproduced in Appendix D.
The implementation of the first iteration was completed
on the 2nd of September 2013. I experienced a lot of trouble getting the
asymmetric I/O and stream multiplexing to work correctly, but since the 2nd of
September, it has been solid.