In bug 404645 we're updating the urlclassifier database faster than the async queue can keep up. As a result, the async queue just keeps growing while it waits to write stuff out to disk. It would be easy enough to throttle the database update to keep this queue small, but the urlclassifier doesn't have a particularly good idea of how big that queue is. I propose: * Choose a "limit" for the async queue, say a few megs * When the async queue goes over this limit, send a "storage-backoff" event to the observer service. * Once the async queue is emptied again, send a "storage-resume" event to the observer service. * add some sort of isRequestingBackoff to mozIStorageService to query the backoff state. * Clients that are writing a lot of data to storage can listen for these events and throttle appropriately.

We no longer do async IO, so this bug is no longer valid.