Arista 7060CX-32S 32x100G Switch

The Arista 7060CX-32S is a 1RU fixed-configuration 100 Gigabit Ethernet spine and leaf switch built on the Broadcom Tomahawk (BCM56960) ASIC. It delivers 3.2 Tbps of non-blocking switching capacity across 32 QSFP28 ports that each support native 100GbE or breakout configurations to 4x25GbE or 4x10GbE, providing flexible density for modern high-performance data center fabrics. A dedicated out-of-band management port, a serial console port, and dual hot-swappable power supplies complement the hardware design, while the front-to-rear (F) airflow variant of the DCS-7060CX-32S-F aligns with standard hot-aisle/cold-aisle deployments.

The switch runs Arista EOS (Extensible Operating System), a single-binary, Linux-based network operating system that presents a consistent CLI and API surface across the entire Arista portfolio. EOS supports OpenConfig, NETCONF/YANG, gNMI, and a rich REST API, enabling full integration with infrastructure-as-code toolchains, streaming telemetry pipelines, and third-party orchestration systems. Features such as ECMP hashing, BGP EVPN, VXLAN data-plane encapsulation, and Arista CloudVision integration make the 7060CX-32S a natural building block for large-scale spine-leaf and Clos fabric architectures.

First introduced in Q4 2016, the 7060CX-32S quickly became a reference platform for enterprises adopting hyperscale-inspired network design principles. Its combination of merchant-silicon determinism, deep buffer visibility, and wire-rate 100G forwarding at 1RU form factor addresses the bandwidth demands of high-frequency east-west traffic generated by distributed applications, GPU compute clusters, and large-scale virtualized workloads.

Ideal for

• Spine layer in a multi-tier BGP-based spine-leaf data center fabric connecting 25G or 100G leaf switches
• 100G interconnect for high-performance compute clusters, including AI/ML training environments requiring sustained all-to-all bandwidth
• VXLAN EVPN overlay fabric underlay, providing the high-bandwidth, low-latency forwarding plane for multi-tenant virtual network overlays
• Collapsed spine/superspine in medium-scale enterprise data centers consolidating east-west and north-south traffic in a single 1RU device
• Network disaggregation pilots where open APIs and EOS programmability replace proprietary management planes with DevOps-aligned automation workflows
• High-density server interconnect using 100G-to-4x25G breakout cabling to maximize port utilization in dense rack environments

Technical specifications

Available from Omnixon Global. Submit an RFQ and our team will confirm configuration and availability for your order.