|
|
DriveNetsProfile date: 2026-08-23 DriveNets: Company Overview and Technical Solutions
1. Executive Summary
DriveNets is a high-scale networking software company that pioneered the disaggregation of core routing and high-performance networking infrastructure. Founded in 2015 by Ido Susan and Hillel Kobrinsky, DriveNets operates globally with major operational hubs in Israel and the United States.
The company transforms traditional monolithic routing and networking hardware into a distributed, cloud-native software model. By decoupling network software from underlying merchant silicon hardware, DriveNets allows telecommunications service providers, hyperscalers, NeoClouds (GPU-as-a-Service providers), and large enterprises to build open, cost-effective, and scalable networking infrastructures. Its technology powers substantial portions of global internet backbone traffic across tier-1 carriers, including AT&T, Comcast, and KDDI.
Continue…
2. Core Architectural Philosophy
DriveNets applies the principles of hyperscale cloud virtualization to carrier-grade routing and artificial intelligence backend fabrics.
- Network Disaggregation: Traditional networking relies on proprietary, chassis-based routers that integrate hardware, ASICs, and operating systems from a single vendor. DriveNets disaggregates these components, running independent software on standard white-box hardware powered by merchant silicon.
- Distributed Disaggregated Chassis (DDC): DriveNets implements standard DDC architecture models aligned with Open Compute Project (OCP) and Telecom Infra Project (TIP) specifications. In a DDC deployment, packet forwarding and fabric elements exist as independent physical white boxes clustered into a single logical entity.
- Independent Scaling: Control planes and data planes scale independently. Control software runs on standard x86 compute instances, while packet processing scales across white-box line cards and fabric elements.
3. Product Portfolio
A. DriveNets Network Cloud (Service Provider Routing)
DriveNets Network Cloud is an end-to-end routing solution built to handle massive scale across carrier networks, from access and aggregation to core routing and peering.
- DriveNets Network Operating System (DNOS): A carrier-grade, distributed routing operating system designed to run as microservices over containerized infrastructure. DNOS delivers comprehensive routing protocols (BGP, IS-IS, OSPF, MPLS, Segment Routing / SRv6), quality of service (QoS), line-rate security filtering, and multi-tenant capabilities across clustered white boxes.
- DriveNets Network Orchestrator (DNOR): An automated management, provisioning, and operational software suite. DNOR provides zero-touch provisioning (ZTP), telemetry collection, software lifecycle management, performance monitoring, and real-time failure recovery for distributed hardware topologies.
- White-Box Hardware Ecosystem: DriveNets certifies and supports standardized open hardware models built by Original Design Manufacturers (ODMs) such as Accton, Edgecore, and UfiSpace, utilizing high-performance merchant silicon (e.g., Broadcom Jericho and Ramon chipsets).
B. DriveNets Network Cloud-AI / AI Fabric
DriveNets AI Fabric provides a full-stack, Ethernet-based networking architecture engineered specifically for AI/ML training and inference workloads. Designed as an open, standard alternative to proprietary InfiniBand fabrics, it optimizes GPU-to-GPU communications and minimizes Job Completion Times (JCT).
- Fabric-Scheduled Ethernet (FSE): Employs cell-based scheduled packet transmission across network fabrics to eliminate packet drops, prevent tail latency spikes, and optimize collective communications libraries (CCL/NCCL).
- Hardware Platforms (DriveNets 2600 Series): High-density network switching platforms tailored for massive GPU clusters:
- DriveNets 2600SL: A fully liquid-cooled switching platform designed for ultra-dense, thermal-efficient AI data centers.
- DriveNets 2601S: An air-cooled switching platform supporting high-radix port breakout configurations (up to 800G and 1.6T interfaces).
- AI Cluster Orchestrator: Software tooling providing cluster-wide benchmarking, network validation, topology discovery, telemetry, and automated link tuning across NICs, drivers, and switches.
4. Key Solutions and Use Cases
- Core and Edge Routing: Replaces legacy, high-cost core chassis routers with modular DDC clusters scaling from single-terabit capacities up to hundreds of terabits per second.
- Peering and Data Center Interconnect (DCI): Enables edge peering points and cross-data-center traffic transport with line-rate encryption, extensive policy controls, and elastic scaling.
- Converged Aggregation: Consolidates multiple service functions (broadband network gateways, wireless backhaul, enterprise VPNs) onto a unified shared compute and routing pool.
- Scale-Up, Scale-Out, and Scale-Across AI Interconnects: Provides ultra-low-latency fabric fabrics spanning inside the rack (Scale-Up), across clustered compute racks within a data center (Scale-Out), and between geographically separated data center facilities up to 100 km apart (Scale-Across).
5. Professional Services and Support
To ensure deployment reliability across mission-critical networks, DriveNets provides:
- Hardware Validation and Certification: Pre-testing of third-party white-box platforms, optics, and cabling.
- Turnkey Deployment and Integration Services: Architecture design, migration planning from legacy chassis routing, and integration with third-party OSS/BSS systems.
- 24/7 Global Technical Support: Carrier-grade service-level agreements (SLAs), proactive monitoring, and dedicated network engineering assistance.
|
|
