Next Generation Wireless Mesh

Mesh networks are NOT all created equal.

We are certified Azalea Networks Partners (http://www.azaleanetworks.com) and can prove that Azalea mesh solutions will provide not only the fastest, most robust and resilient mesh overall, but also one of the least expensive (measured both in terms of initial investment and ongoing operational cost).

First and second generation mesh technologies used one or two radio bridges to build and maintain the mesh network. While some of these systems support layer three, it’s not a native structural element, but an overlay on top of a large bridged network.

Azalea Networks mesh solutions are implemented as true layer three networks – completely routed solutions. Every access point is a fully aware layer three router with multiple distinct wireless interfaces.

For the same reasons we have abandoned the early bridged networks in the wired world, bridged networks are being abandoned in the wireless one. Today, it’s extraordinarily rare to encounter a bridged WAN – these dinosaurs are typically only implemented to service a legacy un-routable protocol.

While bridging was the norm in the early days (early on, it was the only option), it lacks the robustness required for a high-throughput, resilient, secure environment. There are insufficient controls for performance, topology management, security and resiliency.

In a routed network, whether wired or wireless, these attributes may be finely tuned and configured for peak performance in a given environment.

The latest generation technologies use multiple distinct multi-frequency radios as fully routed interfaces and incorporate as many as four radios into an individual AP. Radios used for backbone/backhaul are NEVER shared for access purposes, and traffic between interfaces is routed at layer three. In this environment, no traffic is natively bridged on the network and every packet is routed at layer three at every hop (AP) on the network.

A true layer three architecture like this provides myriad benefits:

• Local traffic stays local. This means that all traffic need not be bounced to the root of the network for forwarding decisions to occur.
• Networks may take advantage of multiple ingress/egress points simultaneously. These multiple attachment points to traditional fiber/copper backbones allow traffic to find the shortest path on and off the network dynamically. This DRAMATICALLY improves performance and allows traffic volume to scale as a factor of the number of egress points, WITHOUT having to deploy multiple distinct meshes (though it’s sometimes desirable to do so for other reasons).
• Resiliency and fast re-convergence. Instead of using Spanning Tree (a protocol which wasn’t even designed for network structuring and re-convergence of this type) a layer three network makes decisions based on the absolute momentary shortest path on a packer by packet basis. The re-convergence of a layer three network is measured in seconds and not minutes like a large spanning-tree network.
• No loss of bandwidth – even after many “hops”. A layer three network can provide consistent bandwidth over many multiple hops, without degradation. A bridged network typically displays significant bandwidth degradation after as few as two hops.
• Security. A layer three network provides the ability to provide per-hop protocol and host level inspection and filtration. Network access can be provided by multiple methodologies, with multiple simultaneous overlays, allowing multi-agency and multi-policy coexistence.

Application data streams like video (the network version of the proverbial “Canary in the Coal Mine”) perform flawlessly over multiple hops and re-converge around network anomalies without rebuilding the hop-by-hop relationships of the entire network. Faster healing means better usability.

In fact, a layer-three mesh can seamlessly support events that would completely cripple other networks – topology splits (where a single network is broken in a way that splits the network into two or more distinct segments that cannot communicate by wireless. This sort of event cripples a bridged network, but a layer-three network can continue operation – and can even communicate between the “island” networks using any wired connectivity that may exist.

It is worth emphasizing that performance in a mesh network has very little to do with the speed of the connection between the client and the Access Point. In fact, backbone bandwidth loss and data propagation methodologies have the most substantial impact on all aspects of performance than do the semantics of different client access layers (e.g.: 802.11a/b/g/n). A well designed 802.11a/b/g network using layer three will outperform any bridged (layer two) mesh network running any client access protocol (including 802.11n).