On December 10, Volta and Fujitsu co-sponsored a webinar hosted by IHS Markit featuring Heidi Adams, Executive Director, Network Infrastructure Research. Our topic was Virtual Routing in 5G Transport. Attendees received a special report authored by Heidi. If you missed it, you watch it on-demand here and download the report, the slides and several other resources.
The turnout and the response were great which shows the interest in issues related to 5G and routing. We had over 20 questions from the audience. We thought it would be useful to share some of those here.
What is the difference between VRFs in a traditional router and your cloud-based virtual routers?
Virtual Route Forwarding virtualizes the routing table. This allows multiple routing and forwarding tables to exist on a device. Thus, a PE router appears as multiple routers to CE routers and IP address space to be reused among multiple domains or customers. Volta’s virtual router runs a separate set of protocols and is a separate management domain. This is essential for applications like RAN sharing each virtual router will be managed by different providers. In addition, switch resources can be managed at the VR level to ensure that service levels are handled correctly. Volta’s virtual routers also support VRFs as a feature.
Can you elaborate on how large one of these disaggregated routers could be? Multi-terabit?
There are two factors that determine the scale of a disaggregated router. First is the capacity of the ASIC which is steadily increasing with new generations. Carrier-grade routing applications require ASICs like Broadcom’s Dune family. The Jericho2 is available and supports 10Tbps so a single white box switch can support up to 10 Tbps. The second factor is how ODM manufacturers use this capacity to build out the number and speed of the ports. This determines how many high-speed ports are available either for a single switch or to build a Clos (spine and leaf) configuration. The devices available today can easily scale into the multi-terabit range very cost-effectively.
What protocols do you use to get the packet forwarding from RIB in the cloud to the FIB in the forwarding device? Is this a proprietary protocol or an open protocol?
With Volta, white box switches serve as a shared forwarding resource. When the RIB is created in the cloud, Volta uses patent-pending technology to consolidate and significantly reduce the size of the RIB when building the FIB used by the switch. As a result, significantly more traffic destinations can be handled in the fast path of the merchant silicon ASIC which minimizes the use of the slower path of the CPU of the switch.
How does 5G impact the last mile?
Enhanced Mobile Broadband (eMBB) is expected provide an alternative to fixed lines for high bandwidth services. Since eMBB will be able to reach gigabit speeds this could be an attractive alternative to in the last mile, particularly for SME locations that have not been served by fiber.
Is the Mobile edge computing expected to be co-located with the DU/CU?
That is one of the models (the Smart CO) that IHS showed. The latency budget will determine how close the computing resources need to be so the actual decision will be application-specific. We see the CO and the PoP (in traditional telco terms) to be the main locations for edge approaches like MEC.