massive disk space leak vs thresh_update

[Matthias Andree]

Tom Allison <tallison at tacocat.net> writes:

> I wonder how much perfomance cost we'll pay for all this transactional 
> benefit.

I haven't benchmarked it because correctness is a few chapters before
speed in my book. However, as we need to write a sequential log
synchronously, rather than many scattered writes to a database (that can
now happen asynchronously), we may require fewer and shorter seeks of
the disk heads.

-- 
Matthias Andree