Case study

University of Patras selects Netdata for infrastructure monitoring

University of Patras’ NAM Group, a 5G testbed facility, selects Netdata to meet infrastructure monitoring criteria for speed, simplicity, and extensibility.

5G Upatras Case Study
Case study on University of Patras leveraging Netdata's monitoring solution

About University of Patras, NAM Group

  • Participant in the €20M project, 5G Verticals Innovation Infrastructure initiative (5G-VINNI).
  • Deploying one of the first 5G-IoT testbeds in Europe.
  • Collaborating alongside with Cisco, Ericsson, Nokia, Telefónica, Altice,  Huawei, and more.

Industry

Academia & research

Simplifying complexity

 

Netdata’s zero-configuration and lightweight deployment worked as an ideal setup for NAM.

Faster, detailed analysis

Usage of Netdata’s per-second granularity  for uncovering root-cause analysis.

Out-of-the-box visuals

Netdata’s auto-configured, rich dashboard assisted NAM’s end-users for troubleshooting their services.

Highly interoperable & extensible

Netdata auto-configuration assisted NAM in integrating with collectors like Prometheus and others.

Share via

Netdata selected to monitor EU’s FIRE 5G testbed

The Network Architectures and Management (NAM) group at the University of Patras has a long history of involvement in European Union research projects, including as part of the Future Internet Research and Experimentation (FIRE) initiative. They have also pioneered a civic engagement service, available for free, and are currently deploying one of the first 5G testbeds in Europe.

The 5G-VINNI project involves deploying an end-to-end 5G facility that can support a wide range of use cases. The project involves a trial, end-to-end deployment of the latest 5G technologies for radio access, backhaul, and core networks while leveraging advanced virtualization technologies and optimization algorithms to power various demanding, industry-specific applications and services. The project is realized in conjunction with a number of different partners, including telecommunication operators ICT equipment vendors, SMEs, European research institutes, and universities.

Christos Tranoris, Research Fellow at NAM, describes the project: “Our infrastructure provides the adequate level of openness to make it possible for vertical industries to test their innovative 5G business cases using ad-hoc network resource control in an end-to-end interoperability framework. The Patras 5G site is a 5G and IoT facility with most of the installed components offered as Open Source, such as Openslice and Open Source MANO.”

Given the complexity of 5G deployments and the need for monitoring up and down the technology stack, the researchers sought a tool that was easy to deploy and could give them immediate visibility into the health and performance of the systems and applications. They identified Netdata as an ideal solution. Netdata is largely known as the go-to monitoring solution for troubleshooting an incident, especially when per-second granularity is crucial to uncover the root causes behind a system or infrastructure incident. However, Netdata’s approach to monitoring is not limited to high granularity, but also features zero configuration and high interoperability, making it easy and lightweight to deploy. This approach led the research team to incorporate Netdata into the pilot deployment.

The deployment is called Patras5G, and is offered to partners for experimentation. It provides all the necessary 5G equipment, both in hardware (such as Intracom’s mmWave backhaul that links the access to the core network) and in software (enabling the E2E deployment of multiple slices of the whole network: Access, Transport and Core).

The backend services are hosted on-premises inside a Linux datacenter. The data center is able to create network services, a collection of VMs that perform the whole range of services that a telecommunication network must provide, from the communication with the 5G antenna that enables the cellular coverage for the devices to the backend services which are responsible for handling the connections and performing the use-case specific activities. These network services are part of a catalog, where they are defined and end-users can choose and activate them.

The architecture defines 3 distinct Virtual Network Function Descriptors (VNFD), namely:

  • VMs that run a regular Netdata Agent and arbitrary services that are required for the use case of the “slice.” These services can range from a video streaming service to a virtual network switch.
  • VMs that run Netdata with a special data collector and are responsible for communicating with the 5G radio station.
  • VMs that run Prometheus and aggregate metrics from every node running Netdata.

Each network service has any number of VMs that run arbitrary services, one VM that is responsible for communicating with the 5G radio station, and one VM that runs Prometheus.

To support this process, there are a number of VMs which run the components of the system (e.g. the orchestrator), and they are logically independent from the rest of the system. The cloud platform offers a total computing power of 300 CPUs, 1TB of RAM, and 50 TB of storage.

When a network service is deployed by the controller, the orchestrator starts configuring and deploying the different VMs. As soon as network service bootstraps, it is ready to serve users via the preconfigured 5G base station, and the stakeholders can access the data either via the distinct Netdata Agents for debugging reasons or via the Prometheus interface for generic visualization.

In essence, there are several virtualized and independent logical networks that use the same physical network infrastructure: the host machines, physical network connection, or even the 5G radio base stations. These groups of VMs can be spawned and terminated on request, creating an on-demand 5G network.

Despite the complexity of the infrastructure, due to the many virtualization layers and the experimental nature, the researchers were able to considerably reduce the deployment time by installing Netdata on both the host machines and inside each VM that is spawned. The Netdata Agent, which is installed directly on the host machines (about five in the datacenter), stream their metrics to a central parent node so they have centralized control over alarms.

“We used Netdata because it was very easy to install, both on the Linux host machines and in the VMs,” said Tranoris. “It offers a great collection of plugins out-of-the-box, which are auto-configured, creating a per-second, detailed, rich dashboard for our end-users in case they want to dive into their services”.

The end users are the stakeholders who request the slice from the system so that they can run their services and enable their end-users—the devices which connect to the 5G network.

The integration with Prometheus was turnkey, thanks to the Netdata Prometheus exporter, enabling the team to focus on the complexity of getting the Grafana dashboard right. The team used Netdata to monitor the entire infrastructure and the services, but used Grafana to visualize select metrics to external stakeholders who don’t need all the available information.

Furthermore, the research team used Netdata’s plugin system to prototype a collector for the 5G radio stations from Amarisoft, using Python as their language of choice. Netdata collectors can easily be developed in either Go or Python thanks to readily available helper functions that Netdata provides. This architecture helped them quickly bring into the fold metrics that are highly specialized in nature but absolutely necessary for their business requirements.

“Another reason for selecting Netdata was that it is well documented, and new custom collectors can be developed in a versatile and rapid manner while satisfying our specific needs,” said Tranoris. “We used this functionality as a medium to showcase the ease of monitoring a 5G base station.”

The way Netdata’s plugin system is structured, allows the programmer to spend less time on the data aggregation and focus on what matters: the organization of the aggregated data into meaningful charts. The researchers were able to quickly create a PoC Netdata collector for Amarisoft 5G base stations (gNodeB) by grabbing the data that Amarisoft makes available via their proprietary API. The collector gathers the information, digests it, and sends the organized data back to the Netdata Agent.

The above image illustrates the sequence diagram of the best-case (sunny day) scenario for the collector.

Below, there is an image of the actual PoC in action. The left half of the image is a high-definition video that is being streamed by using the connectivity offered from the 5G network, while the right-half is the Netdata dashboard, showcasing the charts that are produced by the PoC 5G base radio station collector.

How Netdata monitors 5G

The project is still underway, with pilots being set up all over the city of Patras to be used in various IoT scenarios by both the University and the project participants. The pilots will provide 5G and IoT coverage at locations of interest while the backend will enable the end-to-end, automated deployments of multiple customized slices over the whole network (Access, Transport, Core). 

5G and IoT locations that Netdata monitors