Some of our blogs that use W3 Total Cache are being hammered by search engines that throw 5 to 10 HTTP HEAD requests at the site a second, causing small cloud virtual servers to wedge and get very slow until the backlog is processed. Each HEAD seems to cause WordPress to build the entire page (regardless of whether it's cached? can't tell...), but then a subsequent GET seems to take just as much time as the original HEAD took.
First question: Does (or "could") a HEAD request simply look at the cached copy of an existing cached page and return the appropriate info rather than building the entire page? The .htaccess rewrite rules that W3TC inserts seem as they are checking the cache, but when I observe the time it takes to return a HEAD, it suggests that perhaps the rule isn't working.
Second question: Does HEAD cause W3TC to create a new cached copy (if the old one is expired) that a subsequent GET will use?
If this question is hard to answer, I can look at the code myself to try to figure it out, but I have seen comments elsewhere that WP itself does create an entire page in response to a HEAD (which it probably should) and that searchbots consequently clobber the performance of WP sites when hammering WP sites that don't have caches in place.