I would be for that. And I think that could be done rather cheaply by just determining the top 1000 nodes of all time only once a day or so. If it were plain mod_perl, I would just store the id's of those 1000 in an array at a server start time (before forking) and randomly select from that from memory without ever touching the disk. It would only make startup a little longer, but would not affect runtime in general.
Liz
In reply to Re: Re: Random Top Nodes (nice)
by liz
in thread Random Top Nodes
by demerphq
| For: | Use: | ||
| & | & | ||
| < | < | ||
| > | > | ||
| [ | [ | ||
| ] | ] |