What Service Providers Look for in a SmartNIC

SmartNICs, as you might imagine, are network interface cards (NICs) with a brain. This brain can be a CPU, FPGA, or ASIC chip, but whatever the case, it allows the SmartNIC to offload processing that would otherwise be relegated to the server CPU. In more complete SmartNICs, this brain also offers programmability.

A recent survey conducted by Heavy Reading paints a picture of the current state of the SmartNIC industry, from the point of view of service providers. To better understand what companies are looking for, they asked a variety of questions, from how companies select SmartNICs, to which applications they use them for, and what challenges they are facing.

First, Heavy Reading asked whether companies were already using SmartNICs. About a quarter of respondents have already deployed SmartNICs, with another quarter planning to deploy within the year. It is clear that SmartNICs are a valuable tool to the global telecom operator market and are rapidly gaining popularity.

That said, a SmartNIC is a versatile tool. What exactly do these companies seek to accomplish by deploying SmartNICs in their systems?

Heavy Reading found that 5G RAN acceleration at the edge is one of the top three most common applications for SmartNICs, with 46% of respondents focusing primarily on this application, and an additional 23% considering it as a secondary application.

At Ethernity, we find this very encouraging, as one of our biggest areas of focus is developing FPGA-based SmartNICs for use in edge computing and 5G OpenRAN, including 5G DU vRouter and 5G UPF. Furthermore, most of the functionalities that survey recipients expect SmartNICs to handle are in fact supported by Ethernity SmartNICs. These include:

  • vRouter
  • vSwitch
  • vRAN
  • UPF
  • vBNG

An incredible 64% of respondents indicated that they are seeking vRouter as the functionality they most expect from a SmartNIC! Ethernity’s ACE-NIC SmartNIC family, as opposed to the competition, can provide complete virtual routing via hardware, meeting this demand right now and improving time-to-market for service providers.

Finally, Heavy reading asked its audience what challenges they face when it comes to deploying SmartNICs, as well as what metrics they use to select a SmartNIC provider. A variety of challenges were raised, such as applications and workloads not yet being ready, issues of scalability, systems integration, and maximizing utilization. Price, power consumption, and form factor were also mentioned as possible obstacles.

When it comes to the challenges that companies are concerned about, Ethernity can mitigate most of them: Our ACE-NIC SmartNICs are application-ready, built to scale (thanks to the nature of the ENET FPGA core), and kept at a low price due to Ethernity’s patented ENET technology.

In terms of the metrics that companies use to select SmartNICs, the overwhelming majority answered that a combination of performance and price is key. Here, too, Ethernity can satisfy most service provider needs. The benefits of using an FPGA at the core of your SmartNIC include optimized performance (since FPGAs are much better suited to high traffic throughput than CPUs, due to the FPGAs’ parallel processing capability) and reprogrammability even once the SmartNIC is field deployed (which offers future-proofing and significantly reduced costs).

Ultimately, there are many different uses for SmartNICs and no one vendor will be the best in all cases. Nonetheless, it is encouraging to see confirmation that Ethernity can address so many operator requirements and that their goals are well-aligned with what we provide. And – if nothing else – Heavy Reading’s survey shows that SmartNICs are here to stay.

By Brian Klaff