Having built a couple of systems like this, I didn't find this to be a big problem. You have to organize your data in memory somehow, and that tended to be along lines that made for fast access. Occasionally I'd have to index something, which meant adding a hashtable here and there. It also allowed me to index and organize in ways that SQL doesn't make available.

The area where I still most needed SQL was for ad-hoc querying and reporting. I dreamed of building a relational-to-object mapper, but settled for a) XPath queries against our snapshot files, and b) dumping to an SQL database for reports.