Troubleshooting network bandwidth

Performance issues tend to arise when networks grow unplanned.

Small networks are less likely to encounter problems with bandwidth utilization. This is because a smaller system with few nodes covering a limited area is easy to literally monitor.

Finding bandwidth problems

Predicting bandwidth requirements is relatively straightforward, but performance issues in response to throughput levels can be well below your bandwidth capacity.

Often, what appears to be a capacity limit issue is actually caused by some other factor.

This can be an overloaded cable that serves too many endpoints, resulting in long wait times for media access.

Device failure on a part of the network can also create conditions that appear to be caused by bandwidth limits. D

NS errors and duplication of the IP address can cause blocked processes that obstruct the network.

Discovering bandwidth problems

If you want to make the most of your network bandwidth, you need to address issues caused by other network problems.

When your network grows beyond the distance that you can actually see, you need to employ network tools to restore network visibility.

With monitoring tools that display all your network performance metrics, you can ensure the health of network devices and rule out equipment failures as a source of traffic problems.

Collecting traffic data for every link on a network and keeping that monitor running over time will reveal where the bottlenecks lie.

If your bandwidth is sufficient on paper but insufficient in practice, it might be that uneven demand over time or a congestion point is creating a bottleneck.

Balancing bandwidth demand

Changing the hours when heavy on bandwidth work is done may be enough to maximize your bandwidth performance.

If demand increases periodically you can schedule busy tasks for different periods of the day, or ideally outside of office hours. This should resolve peak loads on your network.

When discovering one or two overloaded links, you can identify wich nodes inside that subnet are generating the traffic, solve that problem by moving those nodes to a common subnet so that their communications do not flood the entire network. Avoid that overloaded link.

If apps or protocols flood the network you can consider moving the provision of those services to the cloud.

Deal with bandwidth costs

In order to stop surges in demand you have to eliminate the traffic that pushes your bandwidth to it’s limit.

Once you have stabilized your network and reduced peak demand below bandwidth capacity, you can initiate a second phase of optimization - to spread the load across the system in order to reduce bandwidth provision.

Better-organized load balancers, caches and subnets dramatically reduce the amount of traffic on your network.

You may currently have links and equipment almost without load while other sections of your infrastructure are overloaded.

Balancing the flow of traffic so that all devices are used will lower peak performance levels on any particular link or switch.

This reduces overall bandwidth requirements and saves money.