Utilization statistics are used across many different devices on our network. In this video, you’ll learn about bandwidth, storage, network device, and wireless channel utilization.
<< Previous: Log Management and GraphingNext: Interface Monitoring >>
As a network professional, we often associate utilization with the network. But there’s so many other places where utilization can have an impact on performance.
Let’s start with a network. When we think utilization, we think bandwidth. How much information can I push through the network? And how much of this network am I using? We often calculate network utilization in the number of bits per second. So you’ll see it referred to as kilobits or megabits or gigabits or terabits, or even higher. And when we start looking at utilization on a network, we find that the utilization is either 0 or 100. We are either using the network, or we’re not using the network. There is no in-between.
So we usually measure network utilization percentages based on how much the network is being used over a certain amount of time. So we’ll say the network was 50% utilized, because during a particular time frame, it was in use half the time.
Whenever we start trying to break out statistics in this way, we need to really understand exactly where those statistics come from. And usually, when we describe a utilization percentage, or we’re describing how many bits per second we’re putting through that particular network, we can get a pretty good idea of just how much that network’s being used.
Another important utilization statistic is one for our storage devices. Obviously, everything grinds to a halt when the storage gets to 100% utilization, meaning we’re using all of the free space on the storage devices that we have inside of our servers. We have to think about this finite resource, because it’s difficult to increase that. When we fill up a storage device, we either have to delete files that are on the storage device– which may not be a very good option– or we need to add another storage device, which, of course, is a hardware installation. And there are costs associated with that.
The storage utilization values can also have wide swings, especially if the application is one that writes a lot of temporary information, and then when you close the application, all of that information is deleted. Sometimes, like on my particular server, you can see not a lot of utilization changes over time. I add some log files, but my website is a relatively static site. And it’s always going to have about the same amount of utilization. As I add more articles and as I add more content it will begin to creep up slowly. So it’s useful to know how this has performed in the past.
Your baselines can be very, very important. If you’re in charge of your servers and your storage facilities, you may want to watch this over time, and get a feel for exactly what’s happening. You can use your baselines to start calculating out some type of projections into the future. That way, you can plan on adding more storage or increasing the maximum amount of storage over the right time frames.
It’s also important to check utilizations of our network devices. We have switches and we have routers and we have firewalls and intrusion prevention systems and so much more. And all of those devices have CPUs. They all have memory. They all have network interfaces. And all of those must work together to increase the amount of performance that we’re going to have.
So we want to know of these network devices how much CPU is being used? Do we have enough memory to perform the functions for that particular device? Those are two very limiting factors. If your CPU goes too high and your memory is not available, then the performance of that device will absolutely suffer. It’s hard to increase the CPU and the memory in some of these devices. They’re often purpose-built appliances. They don’t have an option for increasing the memory or swapping out the CPU for something else. So very often, it requires a forklift upgrade. We have to completely remove that device, and replace it with something which has more capacity.
There may be other utilization statistics that are important, as well. It depends on the device. For instance, a firewall relies very much on how many flows are passing through that device at any particular time. What are the maximum number of connections that device can support? How many new connections are being created? And how many old connections are being removed every second? Those types of utilization metrics will help you understand just how well that particular device is going to perform.
We also have to think about our wireless networks, and how they’re being utilized. A lot of the performance of a wireless network will be based on the interference and other devices using the same frequencies, because there are a limited number of frequencies that we can use for our wireless networks. This is a summary of the wireless networks that I can see from my studio. Some of these networks are mine. Some of these networks are not mine. They are external to the studio, and I have no control over them.
One of the things I noticed when I initially scanned is I see a lot of the networks in my area. The one with PM is the Professor Messer networks. I have a PM network and a PM. For 802.11n, you can see the protocols being used there. And those two SSIDs have a certain frequency, or set of channels associated with them.
One of the things you’ll notice is right at the top, the top three strongest signals are all using and sharing the same frequencies. They’re all on channel 11 which is 2,462 megahertz. So there’s an opportunity here to improve the performance by changing the way that these wireless networks are going to be utilized. So I made a change on my access point. And I modified this PM SSID to, instead of using channel 11, I changed it to use channel 6.
This is the modified one here at the bottom. And you can see the signal-to-noise ratio was 54 decibels originally. Now, it’s gotten much better at 73 decibels. The signal down at negative 35 decibels. Now, it has come up to negative 17. It’s a big difference from where we were before. By removing that interference with the other frequencies, I was able to improve the performance in the overall utilization of my wireless network.