Application Response Time
Original instructions
About us
Overview: ART Computation Parameters
The following diagram displays ART-related events and computation parameters. Refer to the Design Principles of M580 Networks appendix for details.
A: missed input scan
6: CPU input jitter
B: missed output scan
7: operation of application logic (1 scan)
1: input turns ON
8: CPU output jitter
2: CRA drop processing time
9: network delay
3: CRA input request packet interval (RPI) rate
10: network jitter
4: network delay
11: CRA drop processing time
5: network jitter
12: output applied
The ART computation parameters and their maximum values (in milliseconds) are described below:
ID
Parameter
Maximum value (ms)
Description
2
CRA drop process time (CRA_Drop_Process)
4.4
The sum of CRA input scan time and queue delay
3
CRA input RPI (RPI)
–
CPU task. Default = 0.5 * CPU period if MAST is in periodic mode. If MAST is in cyclic mode, the default value is watchdog/4).
4
network input time2 (Network_In_Time)
2.496 (0.078 * 32)
NOTE: The value 2.496 ms is based upon a packet size of 800 bytes and 32 hops1.
The product of (network delay based on I/O packet size) * (the number of hops1 the packet travels). The network delay component can be estimated as follows:
I/O packet size (bytes):
Estimated network delay (μs):
128
26
256
35
400
46
800
78
1200
110
1400
127
5
network input jitter (Network_In_Jitter)
6.436 ((30 * 0.078) + (32 * 0.128))
NOTE: This value is based on a packet size of 800 bytes for RIO drops and 1500 bytes for DIO traffic.
formula: ((number of RIO drops) * (network delay)) + ((number of distributed equipment hops1) * network delay)
6
CPU input jitter (CPU_In_Jitter)
5.41 (1 + (.07 * 63))
CPU input queue delay (owing to RIO drops and DIO traffic)
7/8
CPU scan time (CPU_Scan)
–
This is the user defined Control Expert scan time, which can be either fixed or cyclic.
9
CPU output jitter (CPU_Out_Jitter)
2.17 (1 + (0.07 * 31))
CPU output queue delay.
10
network output time2 (Network_Out_Time)
2.496
See calculation above for Network_In_Time.
11
network output jitter (Network_Out_Jitter)
4.096 (32 * 0.128)
Calculate in the same manner as Network_In_Jitter without I/O frames from RIO drops.
12
CRA drop process time (CRA_Drop_Process)
4.4
The sum of CRA queue delay and output scan time.
1. A hop is a switch that a packet passes through on the path from a source (transmitting) device to a destination (receiving) device. The total number of hops is the number of passthrough switches along the path.
2. Network input and output times may be increased when optical fiber is used.
increase = total length of fiber cables * 0.0034 ms/km
Estimating ART
Using the parameters described in the preceding table, you can compute the maximum estimated ART based on the maximum number of RIO modules and distributed equipment for an application.
The maximum ART value is equal to the sum of values in the Maximum value column. Therefore, the ART calculation for a CPU scan time (CPU_Scan) of 50 ms and an RPI value of 25 ms looks like this:
4.4 + 25 + 2.496 + 6.436 + 5.41 + (2 * 50) + 2.17 + 2.496 + 4.096 + 4.4 = 156.904 ms ART
NOTE: If a cable break occurs or a cable is reconnected on the network, add an additional time period to the above ART calculation to allow for RSTP recovery. The additional time to be added equals: 50 ms + CPU_Scan/2.