Closed shaitan closed 10 years ago
Iteration over eblob (not startup but generic iteration) has the same speed when we iterate using 1 or 16 threads.
Multiple threads were added for cases when iteration callback performs so slow operations, that is is comparable with disk read. There are no such operations in our practice, so let's simplificate.
As much as I like code simplifications, but wouldn't it make startup and sort/defrag much slower?