Following on Jörg Hoh's comment about building cacheable pages in the dispatcher... I'd add that this has to be part of you design strategy. There are some common web practices that will bust the dispatcher cache if you are not careful. One example is urls with query strings params which by default are not cached. You can configure the dispatcher to blacklist params but that only helps if you know what params will be added. However, most large scale commercial web sites will use things like tag mangers and campaign tracking tools that will append their own params. (Whitelisting query params may be the better approach since you will never know what a client will add to the query param when they call your web page). Another practice that front end developers sometimes use is to add query strings to CSS files in order to effectively clear the browser cache when the CSS changes. Again AEM can help you out if you version your clientLibs but your developers have to be aware. I could go on with other examples, but suffice it to say, it's good idea keep track of how much caching your actually getting in production. You might be surprised.
