Mellanox Spectrum Open Ethernet Switches
Industry Leading Performance at All Speeds- 1/10/25/100/200/400GbE
The Mellanox Spectrum family of Ethernet switches are purpose built for the modern software defined datacenter. The Spectrum product family includes a broad portfolio of Top-of-Rack (TOR) and aggregation switches that range from 16 to 128 physical ports, with Ethernet data rates of 1GbE, 10GbE, 25GbE, 40GbE, 50GbE, 100GbE, 200GbE and 400GbE per port. Spectrum Ethernet switches are ideal to build cost-effective and scalable data center network fabrics that can scale from a few nodes to tens-of-thousands of nodes.
Mellanox SN2000 and SN3000 series Ethernet switches can be deployed in a wide range of data center networking solutions including large scale layer-2 and layer-3 cloud designs, overlay based virtualized networks, as well as part of high performance mission critical Ethernet Storage Fabrics or Machine Learning interconnect infrastructure. Utilizing an Open Ethernet architecture allows a choice of operating systems, including Mellanox Onyx, Cumulus Linux, SONiC, Switchdev and others. Additionally, Spectrum switches are integrated with Mellanox NEO. NEO is a powerful network orchestration and automation platform that simplifies network operations and features built-in storage and RDMA workflows.
Mellanox makes fabric management as easy as possible by providing the lowest latency and highest bandwidth. This allows IT managers to deal with serving the company's business needs, while solving typical networking issues such as congestion and the inefficiencies generated by adding unnecessary rules and limitations when the network resources are sufficient.
Benefits
Cloud Native Infrastructure
-
Leaf/Spine architectures that easily scale up to 10K+ nodes in 2 tiers
-
Best in class VXLAN
-
Automation with best of breed tools including Ansible, Chef, Puppet, and SaltStack
-
OpenStack Neutron Integration
-
Hyperconverged Infrastructure Integration (Nutanix)
-
Turnkey Data Center Interconnect solutions with VXLAN
Storage or Machine Learning Interconnect
-
Fair, high bandwidth, low latency and bottleneck free data path
-
Robust RDMA over converged Ethernet (RoCE) transport for NVMe-OF or GPUDirect
-
Mellanox NEO based orchestration and seamless integration with various storage solutions
-
Built-in telemetry with What Just Happened (WJH)
Choice of Software
-
Mellanox Onyx, Cumulus Linux, SwitchDev, Microsoft SONiC, and more
Features
Large layer-2 and layer-3 cloud designs
-
High density wire speed 1/10/25/50/100/200/400GbE ports
-
Consistent and low cut-through latency
-
Flexible shared lookup table
-
64-way ECMP with adaptive routing
Network Virtualization with VXLAN overlays
-
Single pass VXLAN routing and bridging without loopbacks
-
10X better tunnel scale
-
6X better Virtual Tunnel End Point scale
-
VXLAN with RoCE Storage and Machine Learning workloads
Fully Shared Buffer
-
Maximum resources for congested ports to support superior microburst absorption
-
Monolithic buffer provides predictable performance and fair traffic distribution
-
Intelligent congestion management and congestion spread avoidance for ROCE
Telemetry
-
Detailed and contextual telemetry with What Just Happened (WJH) to answer the When, What, Who, Where and Why questions as soon as an issue happens.
-
Hardware accelerated histograms track and summarize queue depths at a sub-microsecond granularity. Avoids false-alerts common to simple watermarks/thresholds
-
Inband Network Telemetry*
Use Cases
Cloud Native Infrastructure
-
Leaf/Spine architectures that easily scale up to 10K+ nodes in 2 tiers
-
Automation with best of breed tools including Ansible, Chef, Puppet, and SaltStack
-
OpenStack Neutron Integration
-
Hyperconverged Infrastructure Integration (Nutanix)
-
Turnkey Data Center Interconnect solutions with VXLAN
Ethernet Storage Fabric
-
Fair, high bandwidth, low latency and bottleneck free data path
-
Robust RDMA over converged Ethernet transport for NVMe-OF
-
Mellanox NEO based orchestration and seamless integration with various storage solutions
-
Built-in visibility/telemetry
Machine Learning Interconnect
-
Fair, high bandwidth, low latency and bottleneck free data path
-
Robust RDMA over converged Ethernet transport for GPUDirect
-
Intelligent end-to-end congestion management to maximize many-to-many communication performance
-
Built-in visibility/telemetry