Virtually Big in Japan, with Rakuten Mobile

Part 1

February 4, 2020

The world has been watching Rakuten Mobile closely for their groundbreaking, next-generation 4G/5G mobile network recently launched and built in under 12 months. In November, Netrounds announced that it will deploy its automated active assurance solution in Rakuten Mobile’s cloud-native network across Japan. This is especially exciting to be a part of as it demonstrates the critical role active service assurance plays in assuring the complex networks of today and meeting the requirements for emerging virtualized and programmable networks, such as 5G, at scale. 

Rakuten is a thought leader in many aspects and demonstrates this by implementing on a national scale across Japan a fully virtualized mobile network, effectively separating the software from the hardware in order to run the network on both the edge and in the core.

Besides implementing virtualization, Rakuten Mobile also believes that all provisioning should be automated for maximum efficiency, speed and flexibility. As the CTO at Netrounds, this is music to my ears. I am happy, excited and proud that Rakuten Mobile has chosen Netrounds to be apart of their revolutionary zero-touch and cloud-native network as this is exactly the type of network Netrounds was built to support. Let's get technical on what this actually means.

One of the challenges with network function virtualization (NFV) is the actual separation of hardware and software. In legacy networks a single vendor provides the hardware with the software bundled together with a datasheet attached stating exactly how it performs under various conditions. In a NFV environment you have one vendor providing hardware and another vendor providing the Virtual Network Functions (VNFs). To add more complexity, you will often have VNFs from different vendors on the same compute node. In the instance of network degradations you may better pinpoint where in the stack issues occur with active service assurance. Being able to test and clearly prove that the hardware and network are performing correctly is more important than ever in a mixed environment. Another key concept for Rakuten Mobile is elasticity; this means scaling up and down relevant VNFs as required depending on network buildout phases, current load and possibly even time of day; this all brings on more challenges compared to legacy networks.

Before instantiating a VNF on a new compute node it is critical to know that it has network reachability, quality and performance as required for it to function as intended. To address this challenge Netrounds implements a Service Activation Test (SAT) sequence that is automatically executed for each new compute node that is going to host vCU VNFs in the virtualized radio access network (vRAN). The Netrounds SAT covers several functional and performance areas, especially towards the virtualized evolved packet core (vEPC), including:

  • Throughput measurement
  • Measurement of per-direction delay, delay variation and packet loss per quality of service (QoS) class
  • MTU validation by sending packets of mixed sizes
  • ICMP ping testing towards various important endpoints
  • Routing topology sectionalization using Path Trace

This is done by first instantiating a Netrounds Test Agent VNF, which performs the SAT. Once the SAT is successfully completed, new VNFs can be instantiated with confidence that the network requirements and dependencies are met.

Once the compute node has had its network connectivity initially tested, the same Netrounds infrastructure is used for persistent 24/7 monitoring where a small amount of data is continuously sent towards the vEPC in each QoS class to provide a real-time notification if a degradation of key KPIs should unexpectedly occur on any compute node. This gives assurance that regardless of what type of network fault occurs, Netrounds will detect if it does indeed affect traffic in the network. Since many parts of a modern network may well be built relying on unmanaged infrastructure (leased lines, SD-WAN, etc) it's essentially impossible to monitor the end-to-end performance with methods like passive device telemetry. The more abstract a logical network is from the physical devices and links, the more valuable it becomes to embed active service assurance.

In addition to continuous active monitoring of all nodes, the Netrounds solution can also be used for on-demand troubleshooting to sectionalize and narrow down network problems for easier fault resolution; this is a huge advantage in many instances, helping to avoid sending a technician on-site to perform troubleshooting. The troubleshooting can be done in a similar approach as the SAT by having the Test Agent VNF automatically execute various tests to help locate the fault.

The Netrounds solution also integrates with the front-haul network provided by Altiostar to consume metrics of latency, jitter and packet loss. This ultimately gives the Rakuten Network Operations Center (NOC) a single Netrounds dashboard that shows metrics of both front-haul and back-haul performance, with no need to jump into different systems; great for troubleshooting!

Stay tuned for the next blog post covering more details on how this is all implemented. In the meantime, you can watch a presentation I gave touching on the Rakuten Mobile landscape by clicking the button below.

Watch Video ►