In order to determine if a space is an AGN, it has to be determined if the space adequately fulfills the minimum requirements. In order to do that performance standards for each component must be established, such that measurements can be taken and compared over time, among different sites. Each of the parts of the AGN have different standards by which to measure, this section tries to establish acceptable ranges for each measurement for each component of the space.
Some aspects of AGNs performance are difficult to measure. For these characteristics heuristics may be used to provide a measurement that is relative to the known possible best measurements. For example, multicast connectivity might be measured on the basis of the fraction of the time bi-directional connectivity is available to a core set of known-working sites; display interactivity might be measured by a combination of processor load and window event frequency. For the measurable aspects of the AGN we provide the following table outlining the standards to be considered an AGN.
The latency and jitter described here are end-to-end for one-way local-network measurements. That means it indicates an acceptable level of performance for capture, encoding, network transmission and reception, decoding and presentation between devices directly connected on a local network, in one direction. These are all impacted by the choice of hardware, such as CPUs, memory, capture and display hardware. The impact of WAN latency is not included, since that is a function of the underlying network paths involved, and on inter-continental scale can overwhelm the encode/decode latency.
Latency impacts the interaction between sites. Longer latencies introduce greater formality in conversations, as each site has to wait to avoid talking over another. While larger latencies are acceptable for broadcast or lecture models, they become uncomfortable for other types of events (e.g., workshops and small group interactions).