Large Databases

[Andrew Hume]

does anyone know of, or can recommend, any UNIX database packages
than can handle large databases? for the purpose of this question,
consider large to mean 10^7 tuples. the keys average 40-60 bytes,
the data averages about 30 bytes. i need real fast lookup by key,
reasonably fast update and plausibly fast traversal of all keys.
i have been using weinbergers compressed B-trees but my b-tree is
now 4 levels deep (thus breaking the rule that all b-trees are 3 deep)
and i have to rebuild the database too often (12 CPU hrs on a 750).

PLEASE mail replies to me directly, i will post a summary to the net
if requested.

	andrew hume
	andrew at research.att.com
	research!andrew