Creating a Resilient Network Architecture via Local Redundancy

How new local redundancy features in Equinix Network Edge help you design a highly resilient network

Mason Harris
Creating a Resilient Network Architecture via Local Redundancy

Network outages lead to lost revenue, decreased employee productivity and reputational damage. According to the Uptime Institute’s Annual Outage Analysis 2023, network and connectivity-related issues are the most common cause of IT outages that affect organizations. And the cost of outages is increasing, with more than two-thirds of all outages costing more than $100,000.[1] Therefore, every organization needs to think seriously about network resiliency.

Network resiliency typically refers to how well your network can provide continuous operation and recover quickly from small failures. Redundancy involves including extra components in your network design that aren’t strictly necessary. Resiliency and redundancy are used interchangeably in the industry.

Redundancy can be further broken down into local redundancy and geo-redundancy. Local redundancy typically refers to designing resiliency into the local network within a metro—that could be in the same data center, campus or city, depending on how your architecture is designed. Geo-redundancy is an architecture that expands on the idea of local redundancy by utilizing multiple metros. It protects against natural disasters that occur in one metro. Companies with complex distributed infrastructures often need to combine locally redundant and geo-redundant design to ensure the availability of their networks for business-critical processes.

Building redundancy into your network is essential for creating resiliency and avoiding network downtime. Designing a network architecture that includes a combination of local and geo-redundancy will help you achieve true resiliency. There are many possible ways to do this, and exactly what it looks like for any given company depends on its requirements. There’s no one-size-fits-all approach, and you’ll need to balance your priorities with your budget. But to ensure business continuity, there are design principles you can follow, as well as numerous tools and capabilities to support redundant network architectures.

How to create a resilient network using Equinix Network Edge

Just as you would with physical networking, you need to design for redundancy in your virtual networking. Equinix Network Edge is a virtual networking service that allows users to deploy virtual network functions (VNFs)—including firewalls, routers, SD-WAN gateways and load balancers—on demand. And we’ve recently built new features into the platform to support resiliency.

Equinix Network Edge customers now have several options for how to create resilient and highly available networks through both device and connection resiliency. All these tools and capabilities can be managed through the Equinix Fabric® portal. Since we talked about how to create a geo-redundant infrastructure in a previous blog post, let’s look more closely now at local redundancy by exploring two new connection resiliency features in Equinix Network Edge that are helping customers design resilient architectures: device link groups (DLGs) and EVP-LAN.

Retailer explores resilient networking possibilities

We’ll consider an example to illustrate how Equinix Network Edge can help customers improve local redundancy in their networks: A large global retailer has a pair of redundant virtual routers in Chicago that it’s using to connect its colocation environment to an application hosted in AWS. This device redundancy helps the company achieve high availability at the device level. However, it wants to redesign its network for resiliency to ensure better business continuity and avoid potential service-impacting events at the connection level. As a result, the retailer wants to focus more on building resilient connections. To do so, it explores DLGs first and then later takes its resilient design a step further with EVP-LAN.

Building resilient device link groups

DLGs connect Equinix Network Edge devices to other Equinix Network Edge devices. The device link service creates a Layer 2, full-mesh LAN between two or more Equinix Network Edge devices and can therefore connect redundant devices together to synchronize them using relevant protocols. A single DLG, however, provides no resiliency. You can achieve resiliency by deploying additional DLGs that connect to the primary and secondary Equinix Fabric networks.

Device links were previously limited to one Equinix Fabric plane. If there were an Equinix Fabric outage, planned or otherwise, the customer would also have an outage. With DLG redundancy, you now have the option to configure resilient device links so there’s no single point of failure.

For the global retailer in our example, one DLG connecting its virtual routers doesn’t provide connection resiliency, so it decided to deploy an additional DLG connecting to the secondary Equinix Fabric network and have both devices connect to the DLGs.

Read more about device link resiliency and check out this video on how to create a redundant DLG:

Creating resilient EVP-LAN connections

An EVP-LAN is another option for creating connection resiliency in Equinix Network Edge. Whereas DLGs only connect Equinix Network Edge devices, EVP-LAN connection allows you to connect Equinix Network Edge devices to other Equinix Network Edge devices as well as to the Equinix Fabric port to which physical hardware is connected. Thus, EVP-LAN allows you to supplement your virtual architecture in Equinix Network Edge with your physical architecture in colocation.

Perhaps you need to connect devices to multiple assets in the same or different Equinix IBX® colocation cages or extend a network service provider connection (BYOC) in a given metro to another Equinix Network Edge device in a different metro. With EVP-LAN, you can create a private multipoint-to-multipoint network and connect to that network from your Equinix Network Edge devices.

After introducing DLGs, the retailer in our example wanted to include its physical colocated infrastructure in Equinix using EVP-LANs. This allows the company to maximize its virtualized services deployment with the physical hardware. It also requires maximum resiliency, and therefore the retailer needed to deploy EVP-LANs that span both the primary and secondary Equinix Fabric networks, enabling it to connect to its physical hardware as well. Also, a recent feature in Equinix Network Edge allows multiple devices in the same metro to be part of the same EVP-LAN.

Read more about EVP-LAN support and watch this video on how to create multipoint to multipoint EVP-LAN connections:

Future-proof your network now

Geo-redundancy will help you in the event of a natural disaster that strikes in one metro, but most organizations also need local redundancy to ensure business continuity. We have continued to build new features into Equinix Network Edge to support our customers’ need for resilient network designs.

DLGs and EVP-LAN are options for supporting local redundancy in your Equinix Network Edge network design. Then, to support redundancy across different geographies, you can replicate the same design in another metro.

As you evaluate your network’s resiliency, you should weigh your budget with your business requirements. For example, what applications or workloads are mission-critical? Do you have others for which an outage wouldn’t have a major business impact? As you think it through, bear in mind that Equinix Network Edge now offers a range of tools and capabilities to help you design a resilient network architecture. And we’re committed to ongoing development to adapt to our customers’ evolving virtual networking needs.

For a deeper dive into designing your Equinix Network Edge network for resiliency, check out my four-part series in Equinix Community: Building Highly Resilient Networks in Network Edge.


[1] Andy Lawrence and Lenny Simon, Annual Outage Analysis 2023, Uptime Institute, 2023.

Avatar photo
Mason Harris Principal Product Manager, Network Edge
Subscribe to the Equinix Blog