With the arrival and increased ramp up of 5G Stand Alone (SA) mobile networks, telecommunications operators have rightfully earned a (very large) seat at the cloud-native table. After all, these are fully containerized networks we are talking about- without any VNF stepping stones left. Nothing gets more cloud-native than that.
However, there is a uniqueness to “telco cloud-native” that doesn’t necessarily exist across the board for all “generic cloud-native” companies. 5G telcos are not only tasked with embracing and deploying fully containerized environments, they have to do so while dealing with a lot of institutional baggage. These companies have been operating MASSIVE global networks built on previous generation, hardware-centric platforms and solutions. The task of shifting all this physical infrastructure into the virtual world of cloud-native is monumental. The task of shifting the mindset of how employees, management, and directors approach technical planning and implementation is even greater.
How visibility has been impacted
Visibility solutions for telcos have historically mirrored the production environments they were put in place to monitor. They have been based on monitoring and replicating packets, and the solutions have been built on physical infrastructure. Copper TAPs, fiber splitters, packet brokers, probes- the list goes on and the options have been plentiful. When the 5G telco working groups responsible for visibility were first tasked to shift their teams and solutions into the containerized world, the options available were no longer as plentiful.
This has resulted in a lot of telcos turning to key partners to solve the challenge of cloud-native visibility- they turned to their vendors providing the actual production network components. These vendors were happy to take on the challenge and began successfully developing and delivering proprietary vTAPs, built to mirror traffic from their production environment components to newer virtualized packet brokers, probes, or directly to analytic solutions. We have discussed in previous posts (found here and here) why this packet-based approach is not the best architectural path forward for visibility in the cloud-native era, however, it is also important to consider the economic ramifications of sticking with vendor vTAPs moving forward.
Licenses, licenses, licenses…and limitations
Relying on vendor provided vTAPs for visibility has proven to be a tough economical stance to take for telcos as their containerized networks continue to grow. We have heard from 5G telcos across the board that these solutions are simply cost prohibitive to maintain. The problem lies primarily in the fact that vendor provided vTAPs only provide insight into that specific vendors’ offerings. If any individual 5G mobile network provider embraces the power of “best of breed” vendor selection that containerized networks have largely unlocked, they now have networks made up of multiple vendors. From a visibility standpoint, they now have networks that include multiple vendor vTAPs, and a sprawling number of licenses to pay for to correctly implement them.
As you can see above, this is a complex environment for any telco to maintain. They have multiple vendors providing vTAPs, all of which they must interact with independently of each other. Some vendor vTAPs may have features the telco is looking for, while others may have them on their roadmap. Some vendors may package their output a certain way, while others do it a bit differently. There is a legitimate concern of lack of commonality across vendor provided vTAPs. Furthermore, all these solutions are packet based, and require a point of presence on every pod that makes up the production component, or Network Function (NF), it is monitoring to replicate data for visibility. On top of that, they are notorious resource hogs. It is no industry secret that packet processing/mirroring/replication is a highly cumbersome task. Simply put, you pay a high resource-tax to wholesale copy and move packets in containerized environments.
The above example is a very simplified version of the enormous complexity this approach introduces into 5G SA networks as they continue to scale, and the enormous license costs that inherently tag along for the ride.
eBPF: from shiny to green
So what is the better alternative when looking outside of vendor vTAPs? At MantisNet, we have embraced eBPF as core to our success in providing a visibility solution that is ready for cloud-native times. Tawon CVF (Containerized Visibility Fabric) is built on eBPF technology and allows operators to simplify the technical approach for establishing 5G visibility, while using a solution that is built for performance. Tawon does not require a point of presence on every single pod (just one agent per node), it is standardized across the board, and it does not require a complete replication of the production network via resource heavy packet processing.
Tawon is a message-based system that delivers a scalable approach to 5G visibility, allowing operators to have a single pane of glass across all the vendors present in their network while dramatically reducing the amount of licenses to purchase.
eBPF-based solutions, such as MantisNet’s Tawon CVF (built specifically for 5G SA network visibility), are no longer the “shiny object in the room” discussed by folks as they whiteboard ideas for how cloud-native is going to change things and what new approaches should be considered. We are now talking about proven solutions that have been deployed in production networks and are making a significant financial impact by reducing complexity and increasing performance.
