SIP Request Retransmit Timers

Summary

Detailed Explanation on how to Configure SIP Request Retransmit Timers (T1, T2, T3) 

Discussion

Customers may want to increase or decrease the SIP request timers depending on what they are seeking to accomplish. This document will explain how each of the SIP timers behave on the SIP page within the Dialogic® Media Gateways. Exercise caution when adjusting these timers. Undesired outcomes, ranging from lengthy SIP retransmits to an increase in traffic across the network, may result. The following timers are specified by the standard RFC3261. 

T1 Time (ms) 

The T1 timer, which is defined in milliseconds, specifies the amount of round trip time (RTT), that the client will attempt to send a SIP Request and expect a resonse. By default, the T1 timer is set to 500ms. T1 timeout is defined as: T1 (500ms)*64. So the amount of time the client would attempt to send the initial INVITE request would be 32 seconds. If this timer is decreased to 200ms, the T1 timeout would be decreased to approximately 12.8 seconds. Once this initial timeout occurs, without a received response, the timeout is doubled (2*T1). If no response is received before the new timeout, the request is retransmitted and doubled to (4*T1). Request transmit retries would look like this: 500ms, 1000ms, 2000ms, 4000ms, 8000ms, and so on, to a maximum amount of time defined by T2. The T1 values allowed by Dialogic® Media Gateways are from 100 – 60000 milliseconds. 

T1 is an estimate of the RTT between the client and server transactions. Although it is not recommended, you may use smaller values of T1 within closed, private networks that do not permit general Internet connection. Also, you may choose a higher T1 value, which can be beneficial if it is known in advance that the RTT is larger (such as on high latency access links). See RFC3261 for reference. T2 Time (ms) 
The T2 timer, which is defined in milliseconds, identifies the maximum retransmit time of SIP request messages (except for INVITE). The default value for T2 is 4000ms. So the retransmitting and doubling of T1 continues until it reaches the T2 timer. The retransmit timing would look like this: 500ms, 1000ms, 2000ms, up to 4000ms. From that point forward, retransmits will go out every 4000ms. The T2 values allowed by the Dialogic® Media Gateways are from 200 – 60000 milliseconds.

T4 Time (ms) 

The T4 timer, which is defined in milliseconds, denotes the amount of time the network will take to clear messages between client and server transactions. The default value is set to 5000ms (as specified by rfc3261). The T4 values allowed by Dialogic® Media Gateways are from 1000 – 60000 milliseconds. 

Open

Use Case 

When using the fault tolerance system on the Dialogic® Media Gateway, the gateway can be triggered to fail-over to the next endpoint more quickly, by decreasing the T1 timer. If you configure it to 100ms, the Dialogic Media Gateway will retry the request for appproximately 6.4 seconds before doubling, hitting the T2 (max) timer, and then failing over. If you want to wait longer, you could increase the amount of T1 time to 1000ms, and it would take about 64 seconds to timeout. 

It is not recommended to decrease the value unless absolutely necessary, as undesired results could occur. Decreasing the SIP timers may generate more network traffic. If these timers need to be altered, use extreme caution and be fully aware of what they do.