Getting your Trinity Audio player ready...
|
Dependence on the network is rapidly increasing, underscoring the need for robust network infrastructure and better connectivity.
While ChatGPT has shown how quickly many parts of life will change with the impact of AI. Organizations are racing to build and train AI models, placing a strain on the networking infrastructure. Looking to strengthen critical infrastructure, we take a look back at a number of milestones for DriveNets, looking forward to continued growth in 2024.
#1 Independent testing confirms DriveNets Network Cloud-AI is highest-performance Ethernet solution for AI
Based on trials with leading hyperscalers and independent testing by Scala Computing, DriveNets Network Cloud-AI Ethernet solution delivers exceptional job completion time (JCT) results at the highest workload performance without being locked to a specific hardware – supporting an open and flexible ecosystem of ASICs, ODMs, and optical vendors. The testing showed that at steady state (i.e., no impairment), DDC architecture provides a 10% improvement vs. Ethernet Clos architecture in terms of job completion time (JCT).
#2 Satisfying customer experience
As part of our ongoing customer support, the DriveNets CS team compiled responses from their customer surveys, sent after resolving cases. Not only was the response rate high, but overwhelming customers responded that they were ‘very satisfied’ with the experience of DriveNets customer support.
From the responses, a customer shared that “DriveNets quickly responds to our support requests, and the intelligent & thoughtful team members create an environment for open and constructive discussions that drive resolution.”
#3 Joining the Ultra Ethernet Consortium
The inclusion of DriveNets in the Ultra Ethernet Consortium reflects the company’s role as a leader and innovator in AI networking solutions. Joining the UEC and Network Cloud-AI’s industry-leading Ethernet-based AI networking performance reinforce DriveNets’ leadership role in the efforts to bring high-performance Ethernet solutions to the industry. DriveNets Network Cloud-AI – introduced in May and now available for order with Broadcom Jericho 3-AI based white boxes – offers the highest-performance AI networking solution available today.
#4 KDDI Deploys DriveNets – the first disaggregated IP infrastructure deployed in Japan
DriveNets Network Cloud is providing carrier-grade peering router connectivity across the KDDI network, enabling KDDI to scale its network and services quickly, while significantly reducing hardware requirements, lowering costs, and accelerating innovation. As the first TIP DDBR-compliant deployment, it is carrying live internet traffic that will start helping KDDI to reduce power consumption by about 46% and rack space by about 40% compared to the traditional routers in KDDI production networks.
“DriveNets Network Cloud enables us to quickly scale our network as needed, while controlling our costs effectively,” said Kenji Kumaki, Ph.D. General Manager and Chief Architect, Technology Strategy & Planning, KDDI Corporation.
#5 Contributing to the Evolution of the OCP DDC standard
DriveNets continues to contribute to the Open Compute Project (OCP) distributed disaggregated chassis (DDC) standard, which is becoming the leading open networking standard. In collaboration with AT&T, UfiSpace HPE and Intel, DriveNets contribution to the Open Compute Project (OCP) DDC specification version 3 (v3) was accepted. This latest version of the standard is based on DriveNets‘ Network Cloud architecture and specifically addresses mobile backhaul infrastructure, as well as Ethernet networking for high-performance AI workloads in large-scale clusters. Some of the highlights of the DDC v3 routing spec include:
- Encouraging increased collaboration and vendor adoption of the base specification for the development of 800Gbps DDC platforms utilizing Broadcom’s J3 and R3 ASICs
- Significantly boosting capacity from 192Tbps to 922Tbps (equivalent to 64 * 14.4Tbps)
- Enhancing port density on line-cards and fabric (port exhaustion occurs before bandwidth exhaustion)
- Ensuring interoperability between existing 192Tbps clusters and new 922Tbps clusters
#6 Evolving TIP’s DDBR and applying it to AI Networking
With hyperscalers feeling that connectivity solutions for GPUs are falling behind, and not performing at required scale, the job completion time (JCT), is negatively affected (the key KPI in AI networking). At FYUZ23, DriveNets’ senior sales engineer, Juan Rodriguez, presented how the DDBR architecture can be purposed for high performance AI networking, offering a solution that scales massively.
#7 Diving into AI Networking
AI took the world by storm early this year. Leading analysts and journalists appraised DriveNets jump into the AI networking sphere. 650 Group’s Alan Weckel, noted how DriveNets solution delivers performance at scale, saying “DriveNets brings this scale with impressive AI benchmarking results to a market that is trying to find an optimal solution. We view their entry as a positive for the industry as vendors stake their expertise in these new network designs.”
In Silverlinings, Diana Goovaerts, recounted how “DriveNets accidentally solved one of the biggest emerging issues in the cloud, unveiling new software which will allow hyperscalers and enterprises alike to connect as many as 32,000 GPUs with good old Ethernet technology to run massive artificial intelligence (AI) workloads.”
#8 Highest performance fabric for AI networks
With Broadcom announcing the delivery of Jericho3-AI, enabling the highest performance fabric for artificial intelligence (AI) networks, DriveNets introduced DriveNets Network Cloud-AI, an innovative artificial intelligence (AI) networking solution designed to maximize the utilization of AI infrastructures and improve the performance of large-scale AI workloads.
“Large-scale training and inference of AI models will benefit from networks that can perform at 100% utilization like DriveNets Network Cloud-AI. Broadcom’s Jericho3-AI delivers an Ethernet network with perfect load balancing and end-to-end congestion management, resulting in significant reduction in job completion time compared to any other alternative,” said Ram Velaga, senior vice president and general manager, Core Switching Group, Broadcom.
#9 Leading service providers embracing network disaggregation
At a panel at MWC23 hosted by DriveNets, industry leaders shared their perspective on the progress for disaggregated networking. As reported by Silverlinings, “AT&T touted the success it has had with the vendor’s [DriveNets] disaggregated core network routing kit. Then, Orange Interim Executive Vice President of International Networks Jean Louis Le Roux, said the operator was trialing DriveNets gear this year, while Cayetano Carbajo, vice president for core transport and services platforms at Telefonica, said his company was considering including DriveNets in a trial as well.”
#10 DriveNets Network Cloud Carries More Than 52% of AT&T’s Core Production Traffic
Kicking off 2023, AT&T shared that they “migrated more than 52% of all our production traffic to our Next-Gen Core routers, which are based on the Broadcom Jericho2, Ramon chips and uses the Distributed Dis-Aggregated Chassis (DDC) design powered by DriveNets Network Cloud DNOS software.”
The accomplishments listed for 2023 demonstrate the industry’s confidence in us, enabling us to expand and continue to radically change the way networks are built.
Bring on 2023!
Download the report
Which Network Architecture Is Right for You