0 1
Who can benefit more from 5G?
Regarding 5G, many mainstream media report on the potential market for personal consumption, such as AR/VR games.
But in fact, as many industry analysts have observed, telecom service providers have long been keenly aware that the future market does not come from individual users, but from those companies participating in the fourth industrial revolution, using large-scale machines. Internet companies with 5G services such as mMTC and ultra-reliable low-latency communications (uRLLC) will become the main beneficiaries of 5G technology.
▲ 3 key 5G core services
0
2How to use 5G?
If companies want to use 5G's key core services, they must carry out true digital transformation. First, do not confuse digital transformation with the existing simple digitalization; second, in order to cope with digital transformation, companies must consider reconfiguring their business processes to use machine-to-machine communication to automate processes, decisions and actions, thereby eliminating human intervention The potentially high cost of operation, while shortening the time period for value realization, makes event data generation closer to the event time. The key points that connect various businesses are data, business rules, decisions, and actions that match them.
Many companies have realized that the speed and low latency of 5G can raise their competitiveness to new heights. However, meeting real-time demands requires fundamental digital transformation and transformation. The definition of real-time is based on two key factors:
“
Latency: How long has passed since the data generated by the device to the value of the data is captured
Workload: What operations need to be completed within the delay time to extract value information from the data and form decisions
Complexity vs latency
As shown in the figure above, for simple aggregation and connection, a true event-driven architecture requires more resources than event-driven stream processing. A real real-time workload requires at least 7 steps: ingestion, storage, aggregation, key performance indicator measurement, detection of abnormal indicators (deviations), decision-making to deal with these complex deviations and appropriate measures (profit, threat, prevention) are based on As a result of the decision-making steps, all of these need to be completed in 10 milliseconds or less.
If you miss the processing time window, you must enter the re-check cycle. Of course, when the data processing capacity cannot keep up, the worst case is that the enterprise infrastructure resources will be completely consumed, and no one will know the existence of these data. This is the data nightmare of the enterprise.
03How does the
system's automatic decision-making
equipment system or human operator's behavior affect the system's automatic decision-making? For real-time processing platforms, it is important not only to process data streams, but also to push the prepared training data to the machine learning layer to digest or create new data models.
Once a new data model is created, you only need to include it in the decision-making process to complete the automatic decision-making of the system. Usually real-time data processing and automatic model update will run in parallel in different operating cycles:
“
A cycle is the cycle from event generation to decision-making action. Gartner calls it a Real-Time Control Loop (Real-Time Control Loop)
The other cycle is the event learning decision cycle, which combines streaming data and big data to update the model and manage false positive or false negative scenario decisions
Polished in the application scenarios of customers and partners, VoltDB was originally a distributed memory database, and has now been transformed into an overall real-time data value management platform that can simultaneously process the above two cycles in a dynamic data flow environment.
VoltDB China Contact: sgao##voltdb.com (please replace "##" with "@")
If you are interested in VoltDB's industrial IoT big data low-latency solution or full life cycle real-time data platform management, welcome private messages and enter our official exchange group.