There is a common notion that bandwidth is cheap, and getting cheaper all the time. This is why the popular solution for many network issues, whether bandwidth related or not, is to just throw more bandwidth at it. Although this may alleviate the symptoms of a problem, it does not address the root cause, and can also mask the warning signs of brewing problems, or even ones that need immediate attention, like an infected client reaching out to everyone else on the network and spiking overall bandwidth usage.
So what is a “bandwidth issue” anyway? For many, a bandwidth issue is simply any condition that causes bandwidth utilization to exceed a certain threshold. For others, the frequency and duration of the event is also taken into account. But without thorough knowledge of the traffic on your network when bandwidth thresholds are exceeded, adding capacity is simply throwing money away.
Determining whether or not you really need more bandwidth is a process which should be carefully implemented before throwing bandwidth at any situation. The four steps in this process are audit, analyze, modify, and measure. Let’s look at each of these steps in a bit more detail.
The main objective in the auditing phase is to categorize your traffic based on your business needs. Businesses run on data, so you need to ensure you have sufficient bandwidth to run your business under all conditions, especially during peak loads. But we also know that usage of any asset seems to expand to fill the available resources, whether that usage is critical or not. Let’s call it network entropy in our case. For the purposes of bandwidth auditing, let’s break traffic down into four categories.
First, there’s business-critical traffic. Without room for this traffic, your business will grind to a halt. Typical types of traffic in this category are email, SaaS (including any cloud-based activity), inter-site WAN, VPN, etc., but traffic types may differ between businesses. For example, a video production house may need to also move very large video files on a routine basis.
Second, there’s business-enabling traffic. If this traffic was unavailable, some employees will be inconvenienced, but your business will not stop. Some examples include social media (in some cases), web surfing for research, etc.
Third is harmless traffic. If this traffic was unavailable there would be no business impact, and the level of traffic is such that it typically does not adversely affect the performance of your network. Examples include personal email, audio streaming, social media (in some cases), etc.
Last is toxic traffic. This traffic hurts your business by consuming large amounts of bandwidth on a routine basis, with no benefits to the business. Unauthorized video streaming, peer to peer sharing, and large data transfers fall into this category.
Knowing your type of traffic is critical when it comes to bandwidth management. If you have toxic traffic, the goal is to eliminate the traffic, not provide more bandwidth for it.
Bandwidth management is not just about the type of traffic. It must also factor in the various media over which the traffic travels, and the type of networking equipment in use, since different media types have different capacities and demands, and equipment can have a variety of capabilities and features. One medium to pay particular attention to is wireless, which is a shared medium that is open to interference, and is also “single user.” At any given time only one user can be sending RF signals over the wireless network, and if the throughput to that user is low it affects all other users on the network. What may look like a wireless bandwidth problem is simply one under-performing wireless client that is generating a significant amount of wireless traffic.
Now that you understand your traffic needs and modes of transmission, it’s time to modify your network usage. Of course your business-critical traffic must remain, but perhaps you can reroute some of this traffic to keep it from more congested network segments. Or perhaps you can block certain types of traffic, like social media for all but your social marketing employees, and of course all toxic traffic. How you modify traffic is an individual decision based on your business requirements and corporate policies, but it is for sure a much less expensive, and in many cases more predictable, approach to bandwidth management.
The measurement phase is all about validation. Did your modifications result in the expected bandwidth savings? Do the changes continue to achieve the same results over time? In general the overall process is an iterative one, because demands, both business-critical and toxic, change over time.
And how are these measurements made? We recommend a packet-based network analysis solution, which will provide all the detail you need measure and troubleshoot just about any type of network condition. It’s only through constant measurement that you can stay on top of your network utilization and ensure that you are truly getting your money’s worth out of your network, and not just succumbing to network entropy.
To learn more about bandwidth management and control, please watch our free OnDemand Webcast.