Dual Uplinks with LACP: Built in Network Redundancy on Bare Metal Servers

Intelligent Traffic Distribution with LACP Hashing

OpenMetal’s LACP configuration is tuned to distribute traffic intelligently across both uplinks using a hashing algorithm designed for optimal flow balancing. Depending on your network design and traffic patterns, this algorithm can distribute packets based on:

• Source and destination IP pairs, ensuring even distribution across flows communicating with multiple external systems.
• Layer 4 port information (e.g., TCP or UDP port numbers), allowing finer-grained balancing of diverse application traffic.
• VLAN tags, which is particularly important for customers deploying multi-tenant OpenStack clouds, where isolating tenant traffic into separate VLANs is a common requirement.

This flexibility allows OpenMetal’s infrastructure to adapt to each customer’s specific workload patterns, ensuring that no single link becomes congested while the other sits idle. At the same time, all of this is presented to the operating system and applications as a single logical interface (commonly known as a bonded interface or LAG interface). From your application’s point of view, the server has one unified network path — even though that path is actively spreading traffic across two distinct physical cables and upstream switch ports.

Seamless Failover at the Network Edge

One of the greatest operational benefits of OpenMetal’s dual uplink design is automatic failover at the data-link layer. If a network cable is accidentally damaged, a transceiver fails, or even an entire upstream switch port malfunctions, OpenMetal’s LACP configuration automatically removes the failed link from the bond. No human intervention is required — the switch and server automatically negotiate the loss of link, and all traffic instantly shifts to the surviving uplink.

This means your critical services continue to operate uninterrupted, with only minor (if any) performance degradation. Instead of 20Gbps aggregate capacity, the server temporarily operates with 10Gbps, but connectivity is preserved, ensuring there is no outage, no dropped traffic, and no need for urgent manual troubleshooting.

A Building Block of OpenMetal’s High Availability Network

At OpenMetal, we understand that the network is often the weakest link when it comes to infrastructure reliability. That’s why dual uplinks and LACP bonding are just part of our multi-layered HA design. These uplinks are further enhanced by:

• Redundant network paths to fully independent “A” and “B” switch fabrics
• Dual power supplies on all switches and core routers
• Automated monitoring that detects and alerts on link degradation in real time

This fully redundant design ensures no single network component — from a NIC to a top-of-rack switch — can disrupt service availability. It’s all part of OpenMetal’s commitment to providing highly available infrastructure-as-a-service, designed from the ground up to support mission-critical workloads.

OpenMetal Bare Metal Servers with Dual Uplinks

All of OpenMetal’s hardware (except for the XS) are equipped with dual 10Gbps uplinks, with an effective total usable bandwidth of 20Gbps per server. Below are the OpenMetal servers with dual uplinks and dual boot drives making them the best building blocks for a highly available infrastructure.