Linden Lab's published their Second Life service quality metrics for April and the results are about as poor as you'd expect -- April was a poor month for the virtual world service. 630,000 user hours were lost to global failures, and that doesn't count the number lost to assorted subsystem failures (as you may recall) were abundant through the period, and still remain a daily occurrence.
As for the asset system itself, Linden Lab is having problems with some third-party hardware and software (Fortune, apparently passes everywhere), though the late-night asset problems that are being experienced each night may be less related to hardware than they are to processing batches that add load to central database systems.
So far May has been considerably smoother than April, but that is not to say that it is anywhere near trouble-free.
