Re: RFC: Constant Database Support

[Joerg Sonnenberger <joerg%britannica.bec.de@localhost>]

On Sat, Mar 06, 2010 at 07:44:05PM +0100, Alistair Crooks wrote:
> 5. in the whole scheme of things, access speed is the thing we need
> to measure, i'd have thought? build speed is important, especially
> at boot time on embedded platforms, but that's a one-off.

The current version of the patch (will provide separate soon) is
slightly optimised for the hot case of multiple accesses (sertvent(1)).
The test case is essentially:
        setservent(a);
        for (j = 0; j < 10; ++j)
                for (i = 0; i < 65536; ++i) {
                        svp = getservbyport(i, NULL);
                        svp = getservbyport(i, "tcp");
                        svp = getservbyport(i, "udp");
                }
        endservent();

                        a = 0           a = 1
old services.db         51.1 / 106      3.6/7.5
new services.cdb        15.0 / 140      0.81/0.84

The difference for the related to trap overhead, but I haven't profiled
it to be sure.

Joerg