At some point, NSD will use a more efficient database. The prototype checks what "kernel" to use repeatedly, for parsing zone files, that will also be the case (i.e. once per invocation of zone_parse
). An application should select the "kernel" to use at startup for consistency.