Oxford picked rizz as its word of the year. Merriam-Webster picked authentic, while noting that rizz, deepfake, and coronation also stood out. In a similar vein, we present the 2023 Network Computing words of the year. These terms entered the network industry’s collective vocabulary or took on new meaning in the last 12 months. Let’s look at why these words are so important.
Tied for first place: GenAI and AI
We'll acknowledge runners-up below, but the winning words of the year are GenAI and AI. ChatGPT burst onto the scene a year ago, and since that time, it has been widely embraced across all industries and in companies of all sizes, bringing AI to the mainstream. There are two consequences of this.
GenAI-assisted IT network managers
First, many IT product vendors and infrastructure providers incorporated GenAI and artificial technology into their network management offerings. A typical example would be to use AI to baseline existing application traffic flows and telemetry data and then trigger alerts when any anomalies occur that may indicate a potential performance problem or security incident. In another application of the technology, many vendors added a GenAI component to their offers to assist human IT managers in doing their jobs better or faster. Examples include:
- Cisco’s introduction of the Cisco AI Assistant for Security. This tool helps IT administrators manage their firewalls to maintain a robust security posture. As we noted when the solution was announced, “the Cisco tool lets an administrator ask firewall security and configuration questions. The tool then uses natural language processing (NLP) and machine learning (ML) to answer those questions in seconds.”
- Fortinet’s introduction of its Fortinet Advisor, which provides incident analysis, remediation guidance, and playbook templates. It is designed to help NetOps and SecOps teams investigate and remediate issues and threats. The advice is delivered to the human operators via natural language prompts.
AI requires new infrastructure
The second consequence of the rapid embracement of artificial intelligence, in general, is that enterprise infrastructures must change. AI workloads are different compared to traditional applications running in corporations. At a minimum, AI needs more and special compute resources, including GPUs and new accelerators like Infrastructure Processing Units (IPUs) and Data Processing Units (DPUs).
As we noted in our coverage throughout the year, many organizations must keep AI efforts on-premises to safeguard the data used to train their models. When that is the case, enterprises find that artificial intelligence workloads need more efficient networking and interconnection technology. Some of the notable efforts to address this issue include:
- The formation of the Ultra Ethernet Consortium, an effort that aims to build a complete Ethernet-based communication stack architecture for AI and high-performance computing (HPC) workloads. Founding members of the group include AMD, Arista, Broadcom, Cisco, Eviden (an Atos Business), HPE, Intel, Meta, and Microsoft.
- Chip-based efforts by Cisco and Broadcom to speed up AI workloads. On the network chip front, we reported on Cisco’s introduction of its Silicon One G200 and G202 ASICs that offer enhanced load balancing and other capabilities to improve the performance of an Ethernet-based network and significantly reduce the completion time to run AI workloads.
- Card-based efforts by NVIDIA with its introduction of the SuperNIC, which the company describes as a “new class of networking accelerator designed to supercharge AI workloads in Ethernet-based networks.” It offers features and capabilities similar to SmartNICs, DPUs, and IPUs.
See also our report: How Data Center Infrastructures Must Change to Support AI
Runners-up for Word of the Year
Two other technologies will likely have an enormous impact in years to come.
First runner-up: Wi-Fi 7
Globally, Wi-Fi networks carry more than half of all Internet traffic, according to the Wi-Fi Alliance. Furthermore, the amount of traffic carried over Wi-Fi networks more than quadrupled between 2018 and 2022 and continues to grow.
As we reported in “Wi-Fi 7 May Be Closer Than You Think,” Wi-Fi 7 builds on the technological advances in Wi-Fi 6 and 6E, adding automated frequency coordination, multi-link operations, and 4K Quadrature Amplitude Modulation (4K QAM). Those technologies allow Wi-Fi 7 access points and routers to offer significantly more performance for enterprise users and the ability to support more users in denser environments. Additionally, Wi-Fi 7 enables “extremely high throughputs, low latency and jitter, and increased reliability to support media-rich, immersive experiences,” according to the Wi-Fi Alliance.
A few vendors announced Wi-Fi 7 products in 2023. However, most believe we’ll see more wide-scale product availability in late 2024.
Second runner-up: NaaS
Network-as-a-service is the latest in a line of as-a-service offerings like Infrastructure- and Platform-as-a-Service (IaaS and PaaS, respectively) that provide connectivity in a subscription model. At its simplest level, rather than installing routers, firewalls, switches, and other gear, an enterprise would provision a network through a simple cloud portal, just as it does for compute and storage services. The enterprise would then rely on a hyperscaler or other service provider to deliver the underlying networking infrastructure to run its applications over.
As we reported in multiple articles throughout the year, NaaS caught the industry's attention. MEF, the industry organization that led the development of carrier Ethernet service standards, held the first Global NaaS Event (GNE) in October. The event brought together network, cloud, and technology providers with the purpose of accelerating NaaS offerings.
At GNE, MEF introduced the NaaS Industry Blueprint designed to help speed the deployment and use of NaaS. The blueprint defines NaaS and proposes primary building blocks of NaaS solutions, including services, automation platforms, ecosystems, and certifications. It incorporates existing MEF service and Lifecycle Service Orchestration (LSO) automation API standards that can be used to build and deliver NaaS services. Additionally, the blueprint presented initial NaaS use cases in the areas of on-demand transport, SD-WAN, and SASE.
A final word on 2023
In 2023, the networking industry remained dynamic. Existing and new applications like IoT, immersive environments, and more keep driving the need for more bandwidth and secure connectivity.
GenAI, Wi-Fi 7, and NaaS became household words (at least to IT managers) and will be driving forces in 2024 and beyond.