DEMO: Measuring Connection Statistics

When we think about the internet and the quality of connection we have to the internet, we often jump to bandwidth, and rightfully so. That's a big measurement on how satisfied our end users are going to be with our network: how much bandwidth are they getting? Not enough bandwidth and it could generate poor user experience. Plenty of bandwidth not only can provide that positive experience, but can compensate for other issues that might exist on the network. However, there are some other measurements that we do have to take into consideration.

Running a Speed Test

I'm on my demo laptop here, and what I'm going to do is open up Chrome, and inside Chrome I'm just going to do a speed test. So I'll type in speed test and hit enter. Chrome has this built-in speed test, or I also like to use Ookla quite a bit as well. Either one is going to work.

So we're going to run a speed test here. With this, it's going to show us how much bandwidth you're getting. This is going to be the download speed, so this is measuring the download speed, and then it's going to measure the upload speed. So how much are we able to take down at any given time, and how much are we able to push up at any time up to the internet? We see here we have 223 megabits per second for the download and 21.6 for the upload.

The other thing that we see here is we have a latency of 13 milliseconds. So that is how long a packet will go out and then it takes for a response to come back. So 13 milliseconds, that's like 0.013 of a second, so we're talking about this is happening really fast. So those are the big concerns that we have when it comes to measurement of our network.

Ping, Jitter and Loss

That's not our only concern though. We do have other measurements that we have to take into consideration as well. So what I'm going to do is click the Windows button here and open up a command window, and I'm going to ping 8.8.8.8. This is one of the Google servers. I'm going to use the -t switch so that way it's continuous.

And we're going to take a look at this. I have packets that are going out and then they're coming back, and it's measuring the time. So I see here 66 milliseconds, 41 milliseconds, 23, 845, 47, 115. So those are my different measurements that I have across this period of time.

So what am I looking for here? Obviously bandwidth is one of the things that I really want to focus on here, the download speed, the upload speed. Also, the latency is really important. But as we can see here, I'm not getting that latency consistently. Sometimes I'm getting 120 milliseconds. Here's 176. Here's 269. So it's inconsistent, and that's what jitter is. Jitter is causing inconsistency.

Something I'm not seeing right now is any kind of loss. I'm not having loss at this point in time. So loss would be something that I would measure if I had actual packets that weren't making it out, and I'd have a certain amount of loss with that.

So I stopped this from capturing here, and we see some information here. The total number of packets that were sent were 182. The total number that we received is 182. So we've got 0% lost; there are zero packets lost here. And then we see the minimum amount of time was 13 milliseconds, the maximum was 1,038, and the average was 83 milliseconds.

Recording the Measurements

I've created this Google spreadsheet, and let's fill in some of this data. We're going to figure out the uptime, bandwidth, latency, loss and jitter with this.

First of all, I'll take a look at the speeds. That's 223 and 21.6. So bandwidth, 223 and 21.6. There we have it.

Next up we've got latency. We could use this right here, but this is a pretty simple measurement right here; I want something a little bit more. So what I did is I took the data that we captured right here, just a snippet of that data, and I put it in right here. Now what we can do is figure out what the average is. I could come down here and say average, which is 150 milliseconds. But what I'm going to do is actually use an equation to create an average out of all of this. And so there's our average of 150. So our average latency is 150 milliseconds, with the maximum being 1 second. That's a thousand milliseconds, so that's one second right there. And we also see a half a second in here; 500 milliseconds is a half a second. So there we have our answer right there, 150. So for our latency, 150 milliseconds is that.

For loss we can take a look here and see that we didn't have any loss. We sent 182 packets and we received 182 packets, so for this it's 0%. I tried to get this to tip over the edge and cause some problems, but I wasn't able to do that. So we have a 0% loss right there.

Next for jitter. Jitter is interesting. Jitter is going to be the variation between our average here and what each one of these are. So I'm going to create a little equation here to figure out what the variation is. I've already done some calculation here, and I'm going to explain how I came up with this. Now, jitter could be measured in several different ways, so I'm going to just show you one simplistic way of doing this. What we have here is, this is our first measurement here, 15 milliseconds, versus our average of 150 milliseconds. So if I take the difference there, it's a difference of 135 milliseconds. Here I have 33 milliseconds, and so it's a difference of 117. So you figure out all of the variations here and then we average it. So we have an average variation of 135 milliseconds, which is plus or minus 135 milliseconds, and that is huge. And then here it's just expressed as a percentage, so 90% variation there, which is really big. So I'm going to express this as a percentage and we're going to say 90% jitter rate. We're going to see if we can improve that.

And then we've got the uptime here, which is how much time we were actually down. So if this is a measurement of like over 10 minutes and we are down for a minute, then we'd have a 90% uptime. For this, it was up the whole time, so we're going to say it's 100% uptime.

Improving the Connection

Now what I'm going to do is see if I can improve this. So this is what we're getting at right now, and it's problematic. This would be like your voice phone calls would not be great with this amount of jitter, and this type of connection can cause problems. So I'm going to see if I can smooth this out by restarting the router and the gateway and see what happens here.

I did a restart on the router, and so you can see that it's cleaned it up a lot now. We're getting a much tighter range of jitter within here. Also, I wanted to scroll up and show you what the downtime looked like, because we had some downtime when we restarted that router. So we have here where we have some ping failures, or some requests timed out, destination host unreachable. So during this time, that is what a downtime looked like in this scenario.

So let's do some new calculations here. First of all we've got the bandwidth, so let's run our speed test again and see what we come up with. So here we have 203 and 21.7. So let's put this in here, 203 and 21.7 for the bandwidth.

Now for the latency, we have an average latency of 14.35. So that is a huge improvement. Here we can see how bandwidth is not the only measurement; this latency made the big difference here.

We can take a look at our loss. We have 10% loss here, so there we have our loss. I'm not going to use that number, because that's including the time that I reset the router. So we'll just say since we've got this up and running, we have 0% loss. So we're going to say zero here.

And then for jitter, we're down to 9.41. 9.41 when it's expressed as a percentage, or a 0.001 milliseconds, is really our average here.

And then the only other thing we have here is the uptime, in which case now we're back to 100% uptime. If we wanted to factor in that time I reset the router, then it just depends on over what period of time. It took about a minute for it to do a restart. So let's say through a 10-minute period, that's a 10% downtime. But you'd probably measure it on days or weeks or months or possibly even years is what you'd measure downtime. So we're just going to say since it's been restarted now, it's back to 100%.