TCP RTT – Initial Estimated Round Trip Time in the RTT Formula

latencyprotocol-theorytcp

I was reading about Round Trip Time (RTT) and then I came across this formula

EstimatedRTT = 0.9 * PreviousEstimatedRTT + 0.1 * Sample RTT

What is the value of PreviousEstimatedRTT when it is calculated for the very first time??Is it assumed to be 0 or any other value??

Best Answer

According to RFC 6928, the SRTT is only calculated with the formula

SRTT <- (1 - alpha) * SRTT + alpha * R'

after two RTT measurements have been made. After the first measurement, SRTT is initilized to the value of this first measurement, ie. SRTT <- R'.

You were close, according to the RFC alpha SHOULD be 1/8 (= 0.125).

Note, in RFC's, SHOULD means "there may exist valid reasons in particular circumstances to ignore a particular item, but the full implications must be understood and carefully weighed before choosing a different course." (RFC 2119)

Related Solutions

TCP Round-Trip Time – Why It Decreases 100-Fold with Larger Message Sizes

I will assume from your questions tag that you are using TCP.

A likely explanation that does require some assumptions about your code is that it's an interaction between Nagle's algorithm (an algorithm designed to prevent unnessacery small packets) and TCP delayed acknowlagements (an algorithm designed to avoid sending unnessacery bare acks when an ack could share a packet with application data).

Lets say your applicaiton sends data to the OS in two (or more) parts, first it sends a header, then it sends the actual data. I don't know for sure that this is what you are doing but it seems the most plausible explanation to me. For small messages the following sequence of events happens.

The client application sends the header to the client OS.
The client OS sends the header to the server.
The client application sends the body to the OS
Since there is unacknowlaged data in-flight Nagle's algorithm kicks in and the OS holds back the data hoping for more.
The server OS receives the header, since it's a single packet it holds back the ack hoping for an application response or a second packet to arrive from the client.
We now have something of a deadlock. The client doesn't want to send out another small packet because there is already unacknowlaged data in-flight. The server doesn't want to send out an ack because there has been no application response.
Eventually a timeout breaks the deadlock.

At larger message sizes the interaction changes. Depending on the exact rules the server uses for delayed acknowlagements there can still sometimes be delays but they don't seem to be too significant in your case.

Possible fixes for this problem include using the TCP_NODELAY socket option and/or reworking your application code to build the complete packet before sending it to the OS in one peice.

TCP Latency – What is the Value of EstimatedRTT at First Time?

It can be any arbitrary value, completely dependent on the scenario. For instance, you may measure an initial RTT of 100ms. Now, the RTT measured for the next received packet/ACK may be say, 80ms. For the exponential weighted moving average model, alpha is generally taken to be equal to 0.125 (of course, that may vary as per the model you desire, but it always lies in the interval (0,1)) and the newly estimated RTT would be:

New RTT = (1-0.125)*100 + 0.125*80
        = 97.5 ms

Once again, if the next measured value of RTT is say 150ms,

New RTT = (1-0.125)*97.5 + 0.125*150
        = 104.0625 ms

Thus, the initial RTT that you measure, which becomes a seed for future values, is completely dependent on the network scenario and its parameters.

Best Answer

Related Solutions

TCP Round-Trip Time – Why It Decreases 100-Fold with Larger Message Sizes

TCP Latency – What is the Value of EstimatedRTT at First Time?

Related Topic