7 Warning Signs Your Network is About to Fail – Part 2
This is not intended to be a comprehensive list of network symptoms, just some of the more common symptoms I see when troubleshooting networks.
1. Slow downloads
Slow downloads may occur when you are trying to download a file from a server on your local area network, a server at the far end of a WAN, via VPN or from the Internet. We are all aware of what this looks like because we click the download link and wait. Whether it is a document, spreadsheet or an image file that we need now or a video that we’re downloading to watch later, although it can be annoying we can at least do something else while we wait. It can still be very frustrating when we have expectations as to how long a download should take and we see from the download timer that it is going to take a lot longer. There are many reasons why a download could take a long time – it may not necessarily be congestion. It could be that the Internet or network protocol being used to download this file actually ‘chokes’ the flow of the data so that it doesn’t cause congestion for users running real time applications, it could also be the speed of the source media the file is being read from or the destination media the file is being written to. There are many reasons for slow downloads, which can make for a difficult time when trying to localise the problem.
I would first have a quick survey of other users to see what they are doing or if possible grab some traces from the WAN router LAN interface to understand the other traffic. Maybe the download issues are due the number of user attempting to download at the same time.
If you have established the slow download is definitely not a local device problem and you suspect a congestion problem, the next step is to find out where that congestion is. There a number of free tools we can use to help us localise the problem including PING and traceroute. The ideal test would be to have access to both ends of the data path and run some file transfer tests from a known device at one end of the data path to a known device at the other end. Unfortunately, this is rarely possible and we generally have to work our way through a complex process of elimination with some assumptions thrown in until we get to the bottom of the problem.
2. Web pages not rendering
Web pages failing to render properly may be an indication of congestion. The symptoms may include: images not loading, a text only version of the web page or a web page that simply appears ‘broken’. It is common to assume that the congestion is on the downlink from the internet. However, the cause could also be a result of congestion on the uplink. A congested uplink may prevent the data responses from your device getting back successfully to the website or to the server issuing the web content. This can cause all the symptoms mentioned above. This type of problem is more common across ADSL broadband services as the uplink can become severely congested by uploads such as Cloud storage synchronisation.
I was investigating a regular Monday morning congestion issue for a customer. The problem occurred at other times but on Monday mornings it was unbearable. I found that several of their marketing team members regularly did work at home over the weekend and last thing Sunday night saved the latest version of their work and shut their laptops. Next morning as they arrived in the office and logged on their laptops started to sync their files up to their Cloud storage monopolising their uplink bandwidth for a couple of hours. A simple temporary fix for this was to leave their laptops on for a while after they saved their work to allow the file synchronization to complete from their home networks. A cheap but effective – albeit temporary – solution.
3. Applications running slow
Applications may run slow for any number of reasons – including as a result of network problems. When an application sends data to a client and the response is delayed or lost there are mechanisms in the protocol to recover the session. However, if the protocol struggles to recover the situation the user may be faced with a frozen screen, or typed characters not appearing, or watching an hourglass or spinning beach ball. These are all potential indications of network issues.
Sometimes applications run slow at certain times of the day. For example, the performance issues may coincide with peak activity such as when the majority of users login, print or log off. It may also be because the user profiles are not set up correctly resulting in excessive amounts of data being exchanged during the login process. I have seen examples of this type of problem causing login times in excess of 90 minutes!
4. Applications hanging
In extreme circumstances when network quality is so poor the application protocols cannot recover from lost data the applications are likely to completely hang. This is generally because a response has failed to arrive within a predetermined time and/or attempts, causing the user to be force disconnected from the application.
5. Poor voice quality
Voice over IP uses a protocol that needs a very stable network to operate effectively. If there is any congestion and the VoIP traffic is not prioritised correctly it can cause excessive and random noise problems, jitter problems or the speech signal may sound like someone is talking through a twangy spring. Poor quality voice is generally an indication of high error rate or loss of packets due to network instability.
One of the challenges with voice traffic is that retransmitting a lost voice packet serves no purpose because voice is real time and the moment is lost. Voice over IP transmission is most efficient when it is using UDP, a protocol with no error recovery mechanism.
If you are suffering with poor voice quality there is a reasonable chance you are suffering with network problems somewhere along the path. This may be due to poor configuration resulting in the voice traffic not being handled correctly, a lack of traffic prioritisation or Quality of Service, or congestion. Never assume that Quality of Service is working just because it is configured. Several years ago, my team investigated widespread voice problems across a large network with different WAN and LAN vendors. My engineers discovered the vendors had used different regimes for their QoS bits so the QoS wasn’t actually working despite the claims from both vendors that everything was configured correctly.
6. Video pixelation
When video traffic is suffering as a result of network problems it is fairly common to see video pixellation – rainbow coloured interference in the video stream. This can be a sign of congestion, network errors or that the network isn’t set up properly for video traffic. It is important to note that video traffic is very demanding and especially over wide area network links and Internet connections and the relentless, streaming video may cause network related problems for other applications if not handled correctly.
7. Video freezing
Extreme cases of poor network performance can cause the video to freeze completely. If it is buffered video this may not necessarily present any problems (other than slow download and start). However, for a video conferencing session this can be very disconcerting and render the video conferencing system unusable. The voice element of the video conferencing stream will always take priority over the video stream to the point where the video conferencing system may not attempt to establish the video stream. If the video conferencing system supports screen sharing, this will also fail to operate.
If you plan on using your network or internet connection for video conferencing the network must be appropriately configured and sized in terms of bandwidth and Quality of Service (where possible) especially over WAN connections and the internet. This is particularly important when you are hosting the video hub or bridge providing the conferencing facilities for several endpoints. The network must be able to support multiple video sessions at this location as all video and audio streams terminate on the bridge.
What can we do?
So, what can we do when we are faced with these potential network problems? First, we have to localise the problem – in terms of application(s), physical network location, virtual network location or a combination. We have to go through a detailed process of elimination to identify where the root cause of the problem lies. Is the problem affecting one application, some applications or all applications? Is the problem confined to the Local Area, Wide Area or Virtual Private Network? Is the problem only related to internet access or other specific remote service? If you are part of a multi-site network, are other sites affected? This initial detective work is very important.
WAN links, VPN links and services accessed via the Internet are more likely to suffer with congestion than Local Area Networks. Local Area Networks are typically much faster than any of these other network elements. Contention is also significantly lower on LANs too.
If the problem you are faced with is only affecting services accessed via the Internet, there could be many causes. The problem could be due to your device set up, your local wireless network, Local Area Network, broadband contention, router performance, equipment configuration, and many more. Unless the problem is specifically related to access to one website or service it can be very difficult to determine where the problem lies.
Wide Area Networks can be a particular challenge especially when the access circuits are significantly slower than the Local Area Network connections. If Wide Area Networks are carrying mission-critical data it is important to ensure the access circuits have sufficient bandwidth to support the applications and traffic flows are prioritised appropriately. When WAN access links are scaled correctly and traffic is prioritised with Quality of Service configured correctly the network traffic will run smoothly.
Need help with your network?
Contact us using the form below…