99
104
Many performance problems only become visible when changing the scaling knobs
100
105
upwards to large trees. On small trees its our baseline performance that drives
101
106
incremental improvements; on large trees its the amount of processing per item
103
108
data to be processed under control. Ideally we can scale in a sublinear fashion
104
109
for all operations, but we MUST NOT scale even linearly for operations that
105
110
invoke a latency multiplier. For example, reading a file on disk requires