I maintain here a SharePoint installation, which is growing, so far one crawl component was sufficient. However now the times to completion start taking to long.
During crawl, the WFEs are not to stressed (not by Network, CPU, RAM), so my guess is, there the bottleneck is not located). The Server doing the crawl luckily was a massive physical box, when we got it. So here, even when the crawl is fully ongoing, none of the components seem to reach over 30% utilization.
My idea is now: Is it a limit we hit in the architecture of that single crawl component? If yes, we could run two on that search server. Has someone tried that and can share experience?