Re: Re: Large Objects

Neil Conway wrote:

> > a BLOB.  Conversely, Unix filesystems store directories as unsorted
> > lists, which are a lot slower to search than the database's
> > structured indexes.

> Wow, can anyone confirm this (with Postgres preferrably)? In talking
> with some developers at my old job, they all agreed that storing large
> pieces of data (1k < x < 16K) was significantly faster on the FS than

I believe he's talking about storing all files in the same directory,
which is simply The Wrong Way for a number of reasons. While saving a
large number of external files, we use a sub-dir structure in the form
/data/f4/d3/12/myfile.bin in order to spread the number of files in a
tree pseudorandomly. This is the same approach used by the Squid
webcache.

--
Alessio F. Bragadini        alessio@albourne.com
APL Financial Services        http://village.albourne.com
Nicosia, Cyprus             phone: +357-2-755750

"It is more complicated than you think"
        -- The Eighth Networking Truth from RFC 1925