| Description | The width in milliseconds of each element in shaping
delay queue, that is, the shaping granularity.
The shaping granularity is only applied to rate limit
algorithm: Token bucket algorithm with shaping. It
controls how accurately the algorithm quantizes the shaping
delay for a rate exceeded flow. If granularity is large, several
shaping delay values will all be quantized to the same element
in the queue resulting in less accurate rate shaping for the flows
in bits/sec. On the other hand, choosing too small granularity
causes more memory to be used for the shaper block, and also
can cost a bit more in runtime overhead.
If the rate limit algorithm is not Token bucket algorithm with
shaping, the value is always na(1) which is not applicable.
Setting to any thing other than na(1) will be rejected.
The enumerations for shaping granularity are:
na(1): shaping granularity is not applied to the current
rate limit algorithm
msec1(2): shaping granularity in 1 milliseconds
msec2(3): shaping granularity in 2 milliseconds
msec4(4): shaping granularity in 4 milliseconds
msec8(5): shaping granularity in 8 milliseconds
msec16(6): shaping granularity in 16 milliseconds
At the downstream, this object is configurable and the
default value is msec4(4). Whenever the downstream rate limit
algorithm is changed to Token bucket algorithm with shaping
from other rate limit algorithm, the value will be reset to the
default value. When the algorithm is Token bucket algorithm
with shaping and set to same algorithm again the value will
not be reset to default value.
At the upstream, this object is not configurable and
it is read-only value. |