There are few areas of networking so problematic, and at the same time so fixable, as network congestion. Understanding the common causes of network congestion can help you detect them, fix them, and keep them from cropping up again.
How to identify network congestion
Network congestion is generally seen by the end-user as “network slow down”, or response times on our computer not being up to par. That dreaded “the network is slow!” or “the internet is slow”, and sometimes even “my computer is slow”. To the end user, all of these problems seem the same. But there’s more to network congestion than this.
There’s a variety of tools in your network management toolbox to help you identify and solve network congestion issues. Let’s dive in.
There are five primary ways we see network congestion:
1. Bandwidth Issues
Probably the most common cause of network congestion is plain old bandwidth. Bandwidth is the maximum rate that data can travel along a given path — that path’s total capacity. When there’s simply not enough bandwidth to handle the amount of traffic you have for a particular network, you’ve got network congestion.
Time for a road analogy. If you have a two-lane highway that’s designed to handle 1000 cars per hour, that’s the highway’s bandwidth. Now, if you try to shove 3000 cars down that same highway (same cars, same weather conditions, same time of day), all the traffic will begin to slow down. This slowdown is congestion. The lack of sufficient bandwidth is the main cause of this congestion.
Latency is the delay in the time it takes for your data packet to get from point A to point B. So, using our same example above about bandwidth: You’re a driver on the highway, and you’re traveling the speed limit. Typical day, typical conditions. But what if all of a sudden you hit rush hour traffic? As we explained above, you’ll have to slow down. You’re slowing down to avoid collisions with the other cars. And so is the car behind you, and so on. So now, what would’ve taken you X amount of time at the speed limit is going to take a lot longer (thanks, traffic). That new slower time is your latency.
Latency normally works hand in hand with bandwidth and other congestion issues. So while latency typically isn’t the cause of network congestion, it’s definitely one of the symptoms, as we’ll see later.
Jitter is variability in delay. Computers, like drivers, like to have their traffic consistent and predictable. And when traffic becomes inconsistent, or unpredictable, it causes variability in delay (jitter), and causes further congestion.
Back to the highway. Not all the extra cars appear on the highway at the same time, and they don’t all exit at the same time. For networks, that could be a computer that starts sending large bursts of traffic on the network, taking up excessive amounts of bandwidth.
Every time the network tries to adjust, the computer changes its traffic patterns. This is how jitter creates congestion. The network support equipment is trying to adjust to the variability and can’t keep up. In order to avoid collisions on the network, your computer will initiate a random back-off, and pause sending any packets for a random period of time, measured in milliseconds. This causes the other transmitters on the network to wait before trying again, increasing congestion as a cascading effect. Which leads to our next culprit.
4. Packet retransmissions
Packet retransmissions are usually a result of the first three congestion issues. If a packet doesn’t get to its destination, or if it arrives damaged, then it must be resent. And this has a way of exacerbating the problem. If you need to send each packet two or more times to reach the destination, you’re increasing traffic congestion without any incremental benefit. It’d be like taking the family on a road trip, but every person takes their own car!
The back-off process, mentioned in relation to jitter, is a severe situation where all packets have to wait for the network to clear before retransmitting. Normally this is due to packet collisions on the network, the result of bad equipment or poor cabling. When packets collide, they’ll use this process and a timer to determine when they can retransmit. This leads to even greater delay and congestion. Just like a collision on a highway, the police have to stop and direct traffic for everyone’s safety.
Causes of network congestion
As with any troubleshooting process, understanding the root cause of network congestion is critical to resolving it. Some are more common than others, but it’s important to recognize all of them if you want to know how to manage them.
Unneeded traffic is a common cause of network congestion. This will show up on networks in different ways. For example, someone streaming Netflix or YouTube videos at work would count as “unneeded” in a lot of situations and causes congestion since video traffic uses a lot of bandwidth.
Other examples would be unsolicited traffic like advertisements or junk VoIP phone calls tying up your bandwidth. You should be able to identify unneeded traffic using your network management console.
A typical business network plays host to a lot of different traffic types. There’s broadcast traffic for network operation, multicast traffic for real-time media streams, and unicast traffic to support the data transfer, voice, and video functions we use every day. What we’d call “business critical” traffic can be any one of these types, but it’s important to understand and prioritize what it is. All traffic is intermixed, and in most networks they’re treated equally by network devices, meaning they all get an equal share of bandwidth. If this is not understood and configured correctly, problems are just a matter of time. This is where Quality of Service (QoS) protocols save the day.
Just like vehicles traveling down the highway, Quality of Service shares the same network, but in an unequal way. Traffic is classified and forwarded based on preset rules. QoS allows you to add a bit of special treatment to your business-critical traffic and real-time applications, getting them through congestion quicker.
Highway time again. If an oversized load is moving down the highway, it’s required to stay in the rightmost lane, where slow traffic is supposed to go. But if the oversized load wanders another lane… excessive traffic congestion is back. On the other hand, a presidential motorcade is going to have a police escort, allowing it to slip past everyone else with the highest priority.
In a smooth-running network, the network manager will decide which types of traffic qualifies as “business-critical”, and reserve the bulk of the bandwidth for it. That could range from voice traffic on a VoIP network in a business call center to the order entry system for a large eCommerce business. The remaining bandwidth is left for other types of traffic. One of the best ways to identify business-critical traffic is with a tool such as Auvik Network Traffic Analysis.
How to fix network congestion issues
Once identified, fixing network congestion is not so difficult. The first and most important step is to understand your network and its traffic flows. If you can’t see it, you can’t fix it. Using a good network management tool is essential and can give you the insight and visibility necessary to fix these problems.
Once you’ve found the problematic application or device, the fix is often as simple as:
- Shutting down the problematic application, if it’s not business-critical.
- Disconnecting the device from the network if it is a rogue, unnecessary device.
- “Gently educating” a user about the proper use of the network.
- Updating the device’s firmware to better react to network congestion.
The exact fix will depend on the cause of the network congestion, but once you have visibility into the cause of the issue, the resolution is more often a business decision than a technical one.
How to prevent future network congestion problems
Prevention is worth the effort and a bit of expense. Once you have your fixes in place, and you have inventoried and mapped your network, you’ll be in a great place to provide ongoing support for future issues. There will always be security issues to deal with. Ongoing network traffic patterns that change. Mergers and acquisitions of other companies and foreign networks.
Your key to preventing network congestion is to put into place constant monitoring and proactive alerting. And responding or planning for these changes. A good tool for this will be Auvik’s automation tools.
Cheat sheet: Know your traffic types, and know what to do with them
Business-critical traffic. You should be able to identify what traffic your business needs and give it priority on your network. Using QoS features found in most network control systems, you should be able to give business-critical traffic a high priority of 4 or 5 using the DSCP or (differentiated services code point) standard. 5 being the highest, and business-critical is normally set at 4. Business critical traffic can also be called low latency data or AF23.
Voice traffic should be set at level 5. The reason we do not set business-critical traffic at 5 is that there is another type of traffic that needs level 5 also known in the DSCP standard as expedited forwarding or EF: voice traffic or other low latency traffic such as delay-sensitive transactions.
Normal business traffic. This is where you would categorize normal business traffic. Email, client-to-server traffic. Network backups. This’ll be the bulk of the traffic on the network, but it’ll never use the bulk of the bandwidth. It’s what is called DF or standard traffic. It does not have priority.
Low-priority data. This would be for your unneeded traffic, basically getting what’s left over. This is called CS1 in the DSCP standard. This would not get any priority and would have to wait for all other types of traffic. It’s also known as best-effort traffic.
Unneeded traffic. The final category needs to be looked for and policed on a regular basis. This traffic needs to be removed from the network to free up bandwidth for the other more important and necessary types. This could be offensive traffic, Netflix, gaming, or worse.
All of these types of unneeded traffic should be located using your network management tools. Once they are cataloged and categorized, they can be filtered to be removed from your network completely.
Your Guide to Selling Managed Network Services
Get templates for network assessment reports, presentations, pricing & more—designed just for MSPs.