104
100
Many performance problems only become visible when changing the scaling knobs
105
101
upwards to large trees. On small trees its our baseline performance that drives
106
102
incremental improvements; on large trees its the amount of processing per item
108
104
data to be processed under control. Ideally we can scale in a sublinear fashion
109
105
for all operations, but we MUST NOT scale even linearly for operations that
110
106
invoke a latency multiplier. For example, reading a file on disk requires