Let’s take a real-time interest rate yield curve as an example. To simplify our discussion without deviating from reality too much, let’s assume a yield curve is a discount factor function, i.e. df(t) where t is the number of days from today. Because a real-time yield curve is constructed from real-time market data, it may have very short life-time during active market hours. As a result, lazy-caching technique is often used to let us only compute those discount factors that have been requested at least once and save us from computing those never-requested discount factors. Quite often we also use a so-called double-checking pattern with the lazy-caching technique to protect us from potential threading related issues.
