March 21, 2016

Network Performance and Capacity Planning: When To Upgrade

Nothing kills network performance more quickly than limited bandwidth. Even worse? For many companies it’s difficult to predict when and how bandwidth overages will occur — is it better to spend more for (potentially) unneeded capacity to avoid this problem, or provision “just enough” and hope for the best? Before you break the budget or take a pass, take a hard look at your network capabilities, throughput and capacity. Here is how you get started.

Bandwidth Bottlenecks

Limited bandwidth causes significant problems for enterprises and educational institutions alike. Consider the case of Brooklyn Technical High School. While students can now learn subjects such as robotics and aerospace engineering, limited bandwidth (shared by students and faculty) makes this a frustrating endeavor. What’s at the root of this service slowdown?

Ultimately, bandwidth problems stem from exponential traffic growth across wired and wireless networks, which are in turn used to support several enterprise technologies:

  • Cloud computing — Public and hybrid cloud sharing and storage services require huge volumes of data sent back and forth across your Internet or private network connection. If you don’t have the bandwidth necessary, these apps struggle and local users feel the strain.
  • CRM/ERP — Increasingly sophisticated CRM and ERP tools don’t simply record consumer and inventory data, but also perform complex analytics to yield actionable insight. Since taxing local servers doesn’t make sense, this task is often outsourced to off-site stacks. The result? Limited bandwidth means slow going.
  • Mobile support — The rise of mobile technologies has led many companies to embrace rather than reject these devices, but it’s easy to forget that phones or tablets connected to your network act like any PC desktop — and are constantly consuming bandwidth.
  • Video/VoIP tools — The days of on-premises communications are numbered, as companies embrace the era of voice-over-IP and video tools. The catch? You need big bandwidth to achieve clear, latency-free calls.

Targeting Traffic

Once you’ve identified a bandwidth shortage, what’s next? One option is stepping on the gas with a blanket bandwidth upgrade — in other words, spending big on a network-wide data pipeline increase. While this solves the short-term problem, it often creates other issues — during non-peak times you don’t need this extra bandwidth, so you’re paying for resources you don’t consume. Another option? Leveraging bandwidth-on-demand. In this scenario you’re able to tap more bandwidth as required — for example, you might need extra room during the holiday season, at a particular time of day or if you’re undergoing an office expansion. And since you’ve got the ability to scale up or down on demand, bandwidth is used (and paid for) only as required.

Start with core network traffic volumes. Before the cloud, before VoIP and other value-added services, how much bandwidth does your company need to get the job done? Next, take a look at the types of traffic moving across your network. These might include peer-to-peer connections, data “transiting” across your network from cloud apps to other third-party servers or packets of digital communications data. It’s also worth looking at sources and destinations: Where does the bulk of your data originate and what’s the destination? The farther the distance and more varied your endpoints, the more bandwidth you’ll need. 


Measure, Plan, Fail 

Once you understand the source of your bandwidth needs and how your traffic moves across the network, it’s time to design a plan to upgrade your infrastructure. Start by measuring your average data transfer rate in relation to your bandwidth, then determine a reasonable amount of overprovision to both meet expected peak demands and provide buffer room if usage outstrips your predictions. This is especially critical if you’re leveraging a cloud provider. Without enough bandwidth available, it won’t be possible for the provider to meet SLA terms in the event of sudden traffic spikes, such as during a natural disaster or in-house server failure.

Next, combine your core network capacity requirements with demand predictions to create a plan for bandwidth upgrading. Before you spend big, however, run active simulations and analyses to see what happens in “extreme” cases. What does failure look like for your network? Is your “worse case” simply application slowdown? Data loss? Total service failure? The key here is staying away from passive utilization measurements that yield good-looking data but can’t be trusted when an emergency strikes. To determine the amount of resting and on-demand bandwidth required, always use clear, current and complete failure data.

Your network needs bandwidth. To keep employees productive and end users happy, you need a network performance and capacity-planning strategy that addresses bottlenecks, targets traffic and relies on worst-case, real-time failure scenarios to drive your bandwidth investments.

Stay Connected

Enter your email address below to receive updates each time we publish new content.