I wrote:
> [for a description of traditional techniques for providing various
> isolation levels, including serializable], Dr. Cahill seemed to
> like (Hellerstein et al., 2007)
If anyone else is interested in this paper, here is additional
information:
Architecture of a Database System. (Joseph M. Hellerstein, Michael
Stonebraker and James Hamilton). Foundations and Trends in Databases
1(2).
http://db.cs.berkeley.edu/papers/fntdb07-architecture.pdf
It covers a lot of ground, not just locking and latching issues.
While the discussion seems very good and very clear, it doesn't get
down to low level locking details -- instead referring people to:
J. Gray, R. A. Lorie, G. R. Putzolu, and I. L. Traiger, *Granularity
of locks and degrees of consistency in a shared data base,* in IFIP
Working Conference on Modelling in Data Base Management Systems, pp.
365*394, 1976.
http://www.seas.upenn.edu/~zives/05s/cis650/papers/granularity-locks.pdf
This 1976 paper is the one which gets down to the nitty gritty
details of how to effectively implement predicate locking with
reasonable performance using index range locks, etc. This is the
paper which should cover most of the questions people raise on this
list where Dr. Cahill has just assumed that the "traditional"
techniques he seeks to improve upon are well known to his audience.
These techniques, or some variation on them, have been implemented
in almost every database I've used or investigated.
-Kevin