New technologies such as cloud, NFV, and SDN are changing traditional networks into hybrid ones. Gartner predicts that cloud infrastructure services will grow 35.9 per cent in 2018, and IDC predicts that SD-WAN adoption will grow at a 40.4 per cent CAGR from 2017 to 2022. These numbers imply a great deal of change in networks, change that introduces significant risk of service disruption from minor – a few inconvenienced users – to major – significant outages visible to customers and executives. Reducing the risk during significant transitions is critical. That’s where network performance management and diagnostics (NPMD) products play a significant role.
Networks are changing faster than ever as they are forced to keep pace with rapidly emerging business requirements, the business-facing side of networking is under continuous pressure to do more, in more places, faster. For example, imagine rolling out a critical new cloud-based financial service without understanding the impact on the rest of the network. Or, what about troubleshooting voice or video call quality without being able to see the path of the packets? Whether you’re monitoring applications in the data centre, cloud, and edge, or gathering critical QoS and end user experience information, NPMD solutions offer continuous insights, service assurance, and control. All of this leads to a deeper understanding of the entire network and reduce the risk of disruption during transitions.
There are four areas on which to focus your NPMD solution to meet these requirements:
1. Establish end-to-end network visibility
Without comprehensive visibility across the entire network, NetOps teams lack the level of situational awareness needed to establish a meaningful performance baseline, and the ability to identify and address potential issues before they impact the business. They simply don’t have all of the pieces of the puzzle. Why? Network issues typically hide behind a veil of uncertainty caused by shadow IT, legacy technology, and software-defined solutions running on the network without proper management or monitoring. Organisations need visibility into all activities throughout the network; not only within on-premise and cloud environments, but remote sites distributed across the world too.
End-to-end visibility means assembling network telemetry, such as NetFlow and SNMP, where it is available, and traffic-based metadata where it isn’t, into a graphical view of network elements and traffic. The goal is to understand performance throughout the network in a manner that gives both a broad view and an opportunity to drill down and get more precise information. This sets the stage for our next requirement.
2. Measure performance baselines
Baselining is an incredibly powerful element of the NPMD process. NetOps teams need to understand how the existing network infrastructure performs using the end-to-end visibility mentioned above and enriched by historical performance data. This way, the impact of network infrastructure transitions can be understood and quantified. Making transitions and optimising your network without baselining is like walking along looking only at your feet; you can’t know whether or not you’re making actual progress.
Baselining must be meaningful, embracing variables such as time of day or regular business practices, with the intent that the baseline accurately models the important parts of your network. Don’t waste time baselining links of little use and poor value, instead look for high-value “choke points” where network performance is visible, and even controlled.
3. Understand and embrace the power of software-defined networking
The more rapidly networks must adapt to the demands on them, the more pressure they are under to transition to SDN (software-defined networking). This greater flexibility can lead to more efficient utilisation and cost savings, but can increase complexity. Transitioning from MPLS to a software-defined wide area network (SD-WAN), for example, can help to ensure optimal application performance across software-defined and virtualised environments, creating better user experiences.
Making this transition successfully requires useful visibility before, during, and after implementation, ensuring that performance is at least maintained and guiding configuration and optimisation. These requirements put demands on NPMD tools to reconcile the inflow of data from both legacy and software-defined network components, and use it to manage the hybrid network with a proactive and holistic approach.
4. Leverage the power of network analytics and machine learning
As network complexity increases, and the time available to adapt to changing requirements (“configuration latency”) decreases, networks are becoming less manageable. “Human-in-the-loop” processes are becoming the bottlenecks, making data analytics and machine learning key to reliable, flexible and future-ready networks.
Machine learning can help NetOps administrators teach the network how to assume control of routine network management tasks, monitor application behaviour, and automatically announce important developments. Properly-leveraged network data analytics provide deeper insights into applications and user behaviour, instrumental when considering anomalies on, or adjustments to, the network. Both machine learning and data analytics can make for a much more reliable network and productive NetOps function, freeing up budget, time and resources that can be used to move forward with other projects.
Technology initiatives such as cloud, SD-WAN and IoT represent a tremendous opportunity for businesses. But without robust, capable network monitoring and diagnostic solutions in place businesses put their IT projects at risk. By establishing comprehensive visibility, measuring network performance baselines, embracing software-defined networking and utilising data analytics and machine learning, businesses can optimise their network and better position themselves to successfully implement new technology. Start focusing on these next-generation NPMD best practices to improve network performance and reduce the risk of inevitable network technology transitions today.
Larry Zulch, EVP and General Manager, Savvius
Image Credit: Billion Photos / Shutterstock