How To: Measure your TCP Session’s Quality!

In this Post, I will explain why and when you measure your ‘TCP Session Quality (Starting with: What is ‘TCP Quality’ anyway?).
How you can analyse it yourself, and I’ll share with you a recent example where I had to use it myself. For your comfort – I will write it down in 3 phases.

Phase 1 – “Use-Cases”:

Why would you ever want to measure TCP for its Quality?
There might be a few reasons to do so:
– Persistent Problematic Session.
– Debugging Long distance\low bandwidth\High-Latency Networks Issues.
– Comparing few different Network Solution for Performance and Resources.

But what is TCP ‘Quality’ anyway?
TCP Session can be measured in a lot of different parameters and ways, and i’ll share the one I find the best over and over again: Using Wireshark (Popular Packet Sniffer) . ( < click it to download)
With Wireshark you can (again,) find so much information – So in this next Phase i’ll narrow it down to the ‘Big Ones’ it’s important for everyone to know!

Phase 2 – Understanding TCP Quality:

RTT (Round Trip Time) \ Session Latency :
Is the amount of time takes a packet (or several) to travel from one end of the session to the other and to be acknowledged.
This Parameter is important because it may indicate a ‘slowness’ in one of the network nodes.

Throughput:
The amount of data being transferred by the session over the network in a specific amount of time.
Most Network links are limited for this parameter. Mostly by the ISP \Corporate Q.O.S (Quality Of Service) \ Network Gear.

Re-Transmissions:
A ‘TCP Re-Transmission’ is a packet sent by one end of the session, if it didn’t get the Acknowledgment in time.
Similar to a ‘Packet Loss’.
High amount of those (more than 10%), Usually indicates a serious Issue and the Session will be very slow and unstable.

How do we measure those with Wireshark?
Very simple, After capturing the traffic and filtering your session (right click > Follow TCP Stream).

For RTT/Throughput click (Statistics>TCP StreamGraph>Round Trip Time Graph/ Throughput Graph) :

Now you’ll see the best looking Graphs you have EVER seen!

For Re-Transmissions :

Filter “tcp.analysis.retransmission”

And you will see number of Displayed packets should be less than 10% from total number of packets.

Phase 3 – Real CloudShare Example:

I promised you a Real recent example, and here it is.
Recently, I wanted to ‘put some numbers’ that show how CloudShare’s feature of ‘Accelerated RDP’ is really much better than regular RDP Session.
(Everyone knows it, feels it when working in CloudShare… But how do you put some accurate numbers into it? how much better is it?)
So guess what i did?

In This next scatter-plot, You will see the Round-Trip time over time for two RDP sessions of 1 minute each. (High number is bad)
Left Graph is ‘CloudShare Accelerated’ RDP Session, Right one is Regular ‘Direct’ Session.

With CloudShare’s Accelerated RDP Session, I was able to cut in half (!!!) the amount of time traffic was getting from my PC to my machine in CloudShare.
My User-Experience was much better when working with ‘Accelerated’ , The look and Feel were much better, and now we know why! 🙂

Hope you have Enjoyed this post, And that it was informative to you.
Now, Try Wireshark for your use case – It is Great!