Hey valve. Its ok to have redundant systems (2 separate UPS systems handling half of the data center each for example), or maybe even an extra data center. After all you have gajillions of bucks going into your hands and can easily afford it. If you don't control the data center, then yell at them to get more redundancy, you are paying them enough. You bragged you make more money than Google per employee, so use a small fraction of it.
Maybe hire some extra server specialists who will be dedicated to just that job. That might also help with your load balancing which is CRAP, the store craps itself everytime a big sale or release hits, and its not the capacity since it never maxes itself on the stats, which means its likely load balancing.
Remember, if your store front isn't working, it prevents people from giving you money. Most businesses should give the income systems #1 priority, such as keeping the storefront up 100% of the time. If the system can't handle the game pre-load requests its a pain, but don't make the storefront part of that.