Random Early Detection (RED) is a congestion control mechanism used in computer networks to manage the flow of data packets. Here are the technical details:
- Purpose and Problem Addressed:
- RED aims to provide early signals to transport protocol congestion control (e.g., TCP) that congestion is imminent.
- The goal is to avoid tail-drop losses (where packets are dropped when the buffer is full) and prevent network congestion.
- Operation:
- RED monitors the average queue size in a router or network component.
- It drops (or marks, when used with Explicit Congestion Notification – ECN) packets based on statistical probabilities.
- When the buffer is almost empty, all incoming packets are accepted.
- As the queue grows, the probability of dropping an incoming packet increases.
- When the buffer is full, the probability reaches 1, and all incoming packets are dropped.
- Fairness and Bursty Traffic:
- RED is fairer than tail drop because it doesn’t bias against bursty traffic that uses only a small portion of the bandwidth.
- The more a host transmits, the more likely its packets are dropped, proportional to the data it has in the queue.
- Avoiding TCP Global Synchronization:
- RED helps prevent TCP global synchronization, where all TCP connections back off simultaneously due to tail-drop losses.
- By dropping packets early, RED avoids this synchronization and promotes more efficient network utilization.
- Variants:
- Weighted RED (WRED): Allows different probabilities for different priorities or queues.
- Adaptive RED (ARED): Adjusts RED aggressiveness based on average queue length oscillations.
RED provides congestion avoidance by preemptively dropping packets before the buffer becomes completely full, promoting fair resource allocation and efficient network operation .
