The Complete Network
A wireless network has four separate and distinct components that come together. Each component can either optimize or bog down – and each must be considered when comparing connectivity offerings. We’ve often heard from clients who were directed to focus on only one aspect of the network – ending up with a slower path in aggregate of all the components.
The 4 Components of a Wireless Network
- A-Z Radio Latency
- Physical Path Latency
- Client-Port to Client-Port Latency
- Queuing and Fragmentation Processes
Understanding What Makes Up the Providers Latency Numbers – Truth vs. Hype
A-Z Radio Latency
- Takes into account the deviation from the geodesic distance (the straightest distance “as the crow flies” between any two points factoring in the Earth’s curvature). Note: The closer the path is to the geodesic distance, the lower the latency.
- The gear itself multiplied by the number of links across the path.
- Often quoted by wireless providers, leaving out the two other very critical latency components: physical path and client-port-to-client-port latencies.
Physical Path Latency
Factors in the A-Z Radio Latency along with the fiber facility (commonly referred to as the “fiber tail”) from the last radio into the data center. This should always be considered in the overall latency.
- With the multitude of towers at varying distances around any data center,
- The fiber facility can make or break a path. However, the final component, client-port through client-port, is often the most often overlooked and neglected part of the network.
Client Port to Client-Port Latency
Physical path latency PLUS the equipment latency associated with the provider’s termination device. This should be the only measurement used as it encompasses the entire network, including endpoint-termination devices and their respective queuing and chopping mechanisms.
This is the latency that Anova SLAs on.
Queuing and Fragmentation Processes that Add Latency
A final component in wireless networks is the processes involved in the network provider’s end point termination devices.
Devices used industry wide currently fragment frames into 64 byte packets in order to transport them across the network. Then on the receiving end, those packets are reassembled into their original frame sizes prior to delivery to the client-port. This procedural process adds latency – a total of at least 1.8us.
At 256 byte frames Anova’s system has 10x lower latency than the competitors.
We do not use IP fragmentation in our devices. Our system transmits full packets only adding .9us latency in processing a 256 byte frame.