Real-time is almost as undefined as telepresence. It can be a matter of mere milliseconds and can easily grow to a few seconds.
In effect, when dealing with real-time communication applications, the question we ask ourselves is what is the maximum latency we can get by with and still provide a reasonable experience to the user.
Here are a few target numbers for real-time. You can argue about the exact values are, but I think I nailed the “categories” quite accurately.
| Application | Maximum latency |
|---|---|
| Media encoding/decoding operations* | 20 milliseconds or less |
| Managing protocol signaling* | 100 milliseconds |
| Voice call | 150 milliseconds |
| Video call | 250 milliseconds |
| Live TV show | 2-3 seconds |
| Streaming video | 10 seconds |
* These two “applications” are actually local in nature and don’t deal with network communications. They are thresholds I’d place on the internal architecture of communication devices.
Next time someone talks to you about real-time, be sure to check which kind of real-time and how fast is his idea of real-time.
