I just ran across this excellent post from this past summer by Five Hole Fanatics (hat tip: Battle of Alberta) on the topic of drafting goalies. I think it does a better job of explaining goalie value than anything I have written on the subject. In short, goaltending is a paradox: it is the most valuable position but most goaltenders have little individual value, and the reason is simply that the supply (decent pro goalies) exceeds the demand (available starting jobs in the league). A lot of teams think they need to sign their backstop to big bucks rather than risk losing him, but unless your goalie is one of the top handful of guys you are probably better off signing somebody else who will give you similar performance for a fraction of the price.
As Matt Fenwick at BoA writes: "A goalie who reliably stops pucks at a rate well above average is worth his weight in gold (proverbially; literally, he's worth much more). One who stops them at a below average rate, however, is worthless."
All the value in goaltending is at the top end. Subpar goaltending has no value at all, and there isn't much value in league average performance either. Just look the league's starting goalies who aren't earning huge paycheques, and then account for the fact that some teams have 1 or 2 quality backup goalies just waiting for their shot, and the fact that there are always guys dominating the AHL or KHL or Swedish or Finnish leagues but either haven't got a look in North America or are blocked by the guys above them in their NHL organizations. Studies of minor-league goalie performance (like this one at Oil Droppings) repeatedly imply that there are a number of minor league late-bloomers who go on to have some NHL success, and these guys can usually be had pretty cheaply. A lot of goalie selection and development is probably luck, and teams are somewhat limited in the guys they can develop because of the number of spots available in their systems. However, all an NHL team needs is two goaltenders, so a team with a good scouting and coaching should be able to find two guys capable of playing at league average or better. And once a team has two of them, then a guy playing 70 games at an average level is not actually helping his team much at all.
To make a long story short, and to give a general piece of advice concerning roster management, teams should just do what the Detroit Red Wings do. As GM and former goalie Ken Holland puts it (via Mc79hockey):
"My feeling is if you can get one of the five or six best goalies in the league you can spend the money. We can’t get into those guys, and the difference between the eighth goalie in the league and the 15th goalie, it’s a big difference in money. It’s not a big difference in performance."
If you can get a true difference-maker (i.e. top-5 goalie, which in my estimation appears to currently be Luongo, Giguere, Brodeur, Lundqvist, or Vokoun) then do it, otherwise save your money because you can probably get real close to replacing your performance at a bargain price.