Connectivity Framework
4: Connectivity Framework Layer
based filtering in the connectivity infrastructure is required. For example, a sensor may generate data at a rate of 1000Hz rate, but a user display may not require data at a rate faster than the display refresh rate of, say, 30Hz.
A consumer’ s desired data rate may change over time or for different data items. By knowing the desired data-rate needs across components, connectivity frameworks can optimize the use of system resources for data distribution. This can result in lower overall system resource footprint and lower system cost.
For IIoT systems, the ability to specify a time-based data subset of interest and automatically optimize the data flows is highly desirable.
4.2.3 PERFORMANCE CONSIDERATIONS
4.2.3.1 REAL-TIME
“ Real time” is more about deterministic response than it is about fast response. Many systems require low average latency, but real-time systems succeed only if they always respond“ on time”. This is the maximum latency, and can be expressed as the average delay plus the variation or jitter. Even a fast server with low average latency can experience large jitter under load. For real-time operation, the latency needs to be predictable( i. e. the jitter should be consistently small).
4.2.3.2 LATENCY AND JITTER VS. THROUGHPUT
Throughput refers to the volume of data distributed per unit time. The throughput demands can vary widely— for example, under load or stress in an emergency situation, there may be a lot more communication compared to normal or steady state operation. Latency and jitter can suffer when throughput demands increase on the connectivity infrastructure without increased capacity. A connectivity framework should be able to meet the latency and jitter requirements for real-time performance as with increasing throughput demand, a core consideration of the quality-of-service function.
For IIoT systems, the latency and jitter vs. throughput tradeoffs should be carefully evaluated, and the limiting factors for throughput and latency should be understood.
4.2.4 SCALABILITY CONSIDERATIONS
4.2.4.1 DATA OBJECTS
When the number of data objects increases, it is no longer practical to send every update to every possible consumer. Connectivity frameworks should support data-object scaling by offering runtime introspection so consumers can choose data objects of interest, and configure producer update distribution to a sparser set currently of interest. A producer can also batch multiple dataobject updates destined for the same consumer to make the data distribution efficient and scalable.
IIC: PUB: G5: V1.0: PB: 20170228- 35-