WISP Network Design Overview

Having the capabilities of being self-healing and highly-available are key for operating and growing a successful ISP network. By intentionally building networks around best practices that support these concepts, operators can have peace of mind knowing that the network supports scaling to new tower sites or adding additional distribution points into the network. SDN can also be tightly integrated into networks that are purpose-built, as opposed to those that are organically grown, which opens up opportunities for network automation and integration with external systems such as CRM, billing systems, and NMS (network management systems).

Our solution involves building a versatile network infrastructure and pairing the network with a redundant virtualization deployment. We believe in simple topologies using industry standard/open protocols — straying away from proprietary and often complex vendor-specific redundancy mechanisms. A basic site-redundancy model is shown below. This model consists of identifying a “Primary Site” and “Secondary Site”. These sites are where the key services and functions of your network live.

If localized resources are desired (in addition or as a replacement to) the centralized (shared) resources mentioned above, the design allows for flexibility by building additional sites in a similar manner as the “Primary” and “Secondary” sites. These sites are called “Distribution” sites. “Distribution” sites can be scaled to the appropriate size as needed. When and where redundancy is needed, “Distribution” sites are a requirement. This way, if a failure of any one service at each site occurs, the availability of the service is maintained at the opposite site.

Some operators may desire an architecture where a failure of a service at a given “Primary” or “Secondary” site does not require traffic to shift to the opposite site. Rather, the desire is for the service to maintain its availability within the “Primary” or “Secondary” site. To accommodate, the below topology shows a “Cluster” model. This model duplicates services and resources at the site level and requires mesh connectivity to “Distribution” sites in order to maintain optimal availability.