Multi-Host Solutions


Mellanox's proven Multi-Host technology enables next generation, scalable Cloud, Web 2.0 and high-performance data centers to design and build new scale-out heterogeneous compute and storage racks, with direct connectivity between compute elements and the network. Significantly improving cost savings, flexibility, and Total Cost of Ownership (TCO), Mellanox's comprehensive multi-host technology provides better power and performance, while achieving maximum data processing and data transfer at minimum capital and operational expenses.

Mellanox's Multi-Host Technology Highlights

Mellanox Multi-Host technology is available today on a wide range of Mellanox ConnectX® 10/25/40/50/100 Gigabit Ethernet and VPI adapters ICs, custom cards, and OCP-based boards, enabling Facebook’s Yosemite platform amongst others. Mellanox's innovative Multi-Host technology works by allowing multiple compute or storage hosts to connect into a single interconnect adapter, by separating the adapter PCIe interface into several independent PCIe interfaces. Each interface connects to a separate host CPU—with no performance degradation.

Reducing Data Center CAPEX and OPEX

Mellanox Multi-Host Technology provides huge cost-cutting benefits to customers. Reducing the number of cables, NICs and switch ports required by four independent servers, from four to only one of each, the technology also slashes switch port management and power usage.

Additional Features & Benefits of Multi-Host Technology

  • Enables IT managers to remotely control the configuration and power state of each host individually; guaranteeing host security and isolation, the management of one host does not affect host traffic performance nor the management of other hosts.
  • Lowering total cost of ownership (TCO), Mellanox Multi-Host uses a single BMC, with independent NC-SI/MCTP management channels for each of the managed hosts.
  • Multi-Host also supports a heterogeneous data center architecture; the various hosts connected to the single adapter can be x86, Power, GPU, Arm or FPGA, thereby removing any limitations in passing data or communicating between compute elements.