Describing network traffic

Knowing what traffic flows through the network matters less than being able to describe it in technical terms. These are the properties that matter.

Traffic characteristics

Latency vs. jitter

Latency (often used interchangeably with delay) is the total time from source sending to destination receiving. It includes propagation delay through the media plus processing time on every device along the path.

Jitter is the variation in latency. Network conditions change, so some packets arrive faster than others.

Application categories

Batch

Bulk data transfer with no human involvement after kickoff. The user starts it and waits. Bandwidth affects how long it takes but isn’t critical: even a slow link finishes eventually.

Examples: FTP and TFTP file transfers, inventory updates, automated file replication. (FTP and TFTP are protocols per their RFCs, but the transfers they enable are typically batch-style workloads.)

Interactive

The user acts, then waits for a response. Human-to-machine. Response time matters more than for batch, but strict guarantees usually aren’t required: with less bandwidth the transaction just takes longer and still completes.

How much response time is acceptable depends on how important the app is to the business.

Examples: a database inquiry, a stock exchange transaction, online shopping.

Real-time

Voice and video, usually human-to-human. Bandwidth is critical and delivery is time-critical. Lost data is not retransmitted, so packet loss directly hurts quality.

Sufficient bandwidth is mandatory, and QoS is used to give this traffic higher priority (VoIP being the classic case).

Examples: voice, video conferencing, live streaming.

Mixed requirements in one application

A single application can carry multiple traffic types. Factory automation is the classic example: sensor readings and alarms need guaranteed delivery and a prescribed response time, while device configuration and commercial data from the same application are not time-critical.