Network Resolutions for 2013: Stopping Network Performance Issues Before They Start

Whether you work for a large corporation or a small mom and pop show, network performance is essential. When users are experiencing problems with their network, they become frustrated and aggravated and it is up to the IT admin or network engineer to solve the problem.

There are many factors that can affect the network performance latency, throughput, packet loss and retransmission. We discuss the technical details behind these issues more thoroughly in our blog post, “Four Factors That Affect Your Network Performance.” For this blog post we are going to take a detailed look at how to best curb these issues. Sometimes they will still exist, but there are ways in which you can prepare your network for some of these issues.

Latency: There are a lot of factors that can contribute to latency, including improperly configured applications and overloaded equipment that are getting in the way of transmission. It’s important to remember that latency is additive: once a packet has been delayed, it’s impossible to make up the lost time. VoIP conversations start to degrade with only 150ms of latency, with a generally accepted limit of 250-300ms. In order to help curb latency from happening with your network, it is essential that you follow the packet path using a technique like multi-segment analysis to determine the specific sources. This will help point you in the right direction for what culprit is causing the latency issue. Finding and eliminating a large latency source will go a long way towards helping your real-time communications.

Throughput: Throughput is the amount of traffic a network can carry at any one time. The higher the bit rate, the faster files transfer, but the higher the cost on long-distance links. Slow throughput can manifest itself as long response times and unhappy users. The best way to avoid a throughput problem is to track your bandwidth usage compared to your available bandwidth. Look at your traffic to determine if there are bandwidth hogs, either individual users or misconfigured systems. It’s also a great idea to track bandwidth usage over time, to determine when you’ll need to upgrade your throughput, so you can avoid packet loss.

Packet Loss: Glitches, errors, or network overloading might result in the loss of data packets. Sometimes routers or switches may shed traffic intentionally to maintain overall network performance or to enforce a particular service level. In order to avoid packet loss with your network you must have continued monitoring in place and plan for the point when your business needs exceed your traffic capacity. Packet loss can also occur on “noisy” links, due to bad cabling or failing hardware. Packet loss is relatively easy to spot in modern networks, as many routers count the number of packets they drop. It’s also straightforward to find packet loss using multi-segment analysis along the packet path. While losing the occasional packet won’t make a large difference in your users’ perception of throughput, lost packets in “reliable delivery” protocols like TCP lead to retransmissions.

Retransmission: When packet loss does occur, those lost packets are retransmitted. This retransmission process can cause two delays; one from re-sending the data, and the second delay on the receiver, waiting until the data is received in the correct order before forwarding it up the protocol stack. TCP is designed to make sure that transmitted data is reliably received, but it’s tuned to reduce overall bandwidth usage, so it will wait until it’s received 2-3 messages that data is missing before it retransmits. Then, it will reduce its send rate, and slowly speed back up. Situations causing retransmissions can therefore add a surprising amount of delay into what should otherwise be speedy downloads and transactions. In order to help stop persistent retransmissions, look for lost packets, either data being sent or acknowledgements coming back. Since TCP is based on internal timers,

another less common retransmission source is synchronization problems due to excessive jitter, which is variation in the latency between endpoints.

There are many more reasons that you could be experiencing network performance issues, but these four are some of the more common issues. No matter how well you plan your WLAN or LAN, these network performance issues can and will occur. However, if you plan in advanced and know what to look for, it will be a lot easier to fix problems.

One thought on “Network Resolutions for 2013: Stopping Network Performance Issues Before They Start

Leave a Reply