Forward thinking Telecom Operators are evolving – transitioning their network operations from a multitude of physical devices to a virtualized infrastructure capable of running virtual network functions across a shared pool of resources.
Network analytics must also evolve:
- to gather key data from all layers of dynamic virtualized infrastructures
- perform real-time analytics and anomaly detection
- respond appropriately and automatically to fix issues
- all using a similarly flexible, extensible and scalable architecture.
The key to success is collecting and correlating data from VNFs, VMs, CPUs, Memory, Storage and Network devices as well as being able to act on analytics results in real time.
Today, we would like to introduce IDIA intelligent data integration and analytics, providing modern network analytics from modern telcos. Forward thinking telecom operators are evolving, transitioning their network operations from a multitude of physical devices to virtualized infrastructures capable of running virtual network functions across a shared pool of resources. Network analytics must also be evolved to gather key data from all layers of dynamic virtualized infrastructures to perform real time analytics and anomaly detection to respond appropriately and automatically to fix issues and all using a similarly flexible, extensible and scalable architecture. The key to success is collecting and correlating data from VNS, VMs, CPU, memory storage and network devices, as well as being able to act on analytics results in real time. And that’s where IDIA comes in. Intelligent data integration and analytics for modern telco operators, IDIA enables you to collect data from all NFV layers including VNFs, VNs, VMs and hardware, and to correlate across layers, detector nominees, perform analytics and make predictions.
It can also integrate with machine learning and AI for advanced analytics and modeling and delay you to act immediately on issues by controlling deployment of NFV components. In this demo you’ll see collection of data across NFV layers, real time analytics and anomaly detection, real time visualization of all data and alerting or triggering of external actions. The entry into the application is a global site map showing the status of all data centers worldwide and any important alerts that need immediate attention. Drilling down to a particular site shows the status across an NFV layers. You can also drill down further into problem areas, for example, CPU, which you can then further investigate. You can also see the behavior of various network devices and look for spikes or other issues in network usage. You can investigate the memory usage of different VM instances and how much IO is used by different virtual machines.
There are also VNF level analytics models in this case, analyzing connection requests and failures. Three auto regression models are being tested to identify the best fit of the models to the data basic or to regression or an RR model. Aggressive moving average or armor and auto regressive, integrated, moving average or AREMA. In this case, AR and AREMA of the best fits to our data and can be used for predictions. Nothing in the dashboard is hard coded. It was all built using a dashboard editor. Each visualization is powered by a query to the back end. And configured through simple properties enabling you to rapidly provide the best visualization of your data. The dashboards are powered by data flow applications that collect and process the raw data. This is what the raw data looks like. Sub flows are used to separately processed the physical and VM data with all of the processing done through TQL SQL like query language filtering and transforming the data as necessary and aggregating it to calculate the KPIs. The application can be modified as necessary to process the data according to your particular business needs. This is the flow that calculates the CPU statistics and generates alerts for any issues. A separate application is used to perform anomaly detection with flows for each of the auto regression algorithms. This is the basic auto regression flow. This is the more complex ARMA calculation and this flow implements the AREMA Algorithm, which you saw previously in this dashboard page. Talk to a Striim expert to learn more about IDIA and subscribe to our channel for more videos like this one.