Q: How can we get management buy-in for enterprise deployment of a wireless mesh architecture?
- Raj T., Atlanta

The Wizards gaze into their crystal ball and respond:

Leonid Kalika, Strix Systems

  1. Reliability. The self-tuning and self-healing features of mesh systems means they are inherently more reliable than a wired or partially wireless approach. Should an access point or node fail, mesh systems will route around those failures, ensuring that users are not impacted. Also, mesh systems will typically tune themselves for optimal performance, without IT intervention, so the networks consistently operate optimally.
  2. Scalability. Mesh networking systems are more scalable than other networking systems. Should an IT manager decide that he needs to add additional coverage areas, or add density to existing coverage areas, it is as simple as adding a new access point or node to the existing network. Often this means powering the node on, accepting it into the network and applying the network configuration, and the mesh network goes to work.
  3. Performance. When designing a mesh network, or any wireless network for that matter, it is important to determine the number of users on the network and the bandwidth requirements for those users. As with any wireless technology, there are throughput limitations based on the type of technology. A well-designed mesh system that can provide inter-node link rates of 108 Mbit/s and 54 Mbit/s link rate to clients can provide all the bandwidth required for nearly all the applications used in most enterprises.
  4. Manageability. This goes without saying, but whatever networking system is installed must be not only manageable, but easily manageable. Mesh networking systems self-configure, self-tune, and self-heal, so IT intervention is minimised, but when there is a requirement for user intervention the entire system should be manageable from a single point or interface. The mesh wireless deployment should be a product that was designed as a mesh networking system, not just a collection of access points.
  5. Security. Make sure you select a system that is secure both in the mesh backbone as well as to the client devices. Ideally it will support various authentication (ACL, 802.1x, etc.) and encryption (WEP, TKIP, AES) schemes simultaneously, as well as have some type of monitoring built in.
  6. Cost. This is usually the biggest issue. Mesh networking systems can de deployed for much less than fully wired, or the less wired approach of conventional access points and switches. After initial the initial installation there is always adds, moves, and changes.

[So who is Strix? A wireless mesh vendor recently arrived in Europe - Editor]

Dan Simone, Trapeze Networks
Mesh architectures are primarily targeted at networks where backhaul is expensive to obtain. So rather than rely on carrier circuits or maybe Ethernet to backhaul traffic from wireless nodes, the wireless network itself is used.

Cost justifying such a network involves two things. First, the cost of equipment to build a mesh network with egress (more than one trunk may be required for egress) to get to the Internet or a corporate backbone network solution must be compared to the cost of building a network with traditional access points, plus the cost of a network to backhaul the traffic. Some investment must be made in doing a network design, as mesh architectures are extremely sensitive to traffic patterns. The second thing is whether the performance that can be delivered with a mesh network is acceptable.

Remember that mesh nodes must not only support user traffic but also switch transit traffic, so less bandwidth is available per node. If performance is a critical consideration, then more mesh nodes may be necessary and they typically sell at a substantial premium. Remember that adding more nodes to the mesh means more bandwidth dedicated to backhaul; the network topology is critical in a mesh.