Solve data center network congestion with software-defined automation


Many people seem to have a love/hate relationship with the rerouting feature on Google Maps. My personal view? I love it! On a recent multi-state road trip, it saved me almost two hours by rerouting me around two different cities, successfully avoiding rush-hour traffic. Heavy traffic congestion during rush hour is frustrating, so I’m all for any technology that helps minimize it.

Networks in today’s modern data centers are increasingly prone to this same kind of congestion. Your network is like a highway system your data traverses. Similar to rush-hour traffic, the growing influx and variety of modern applications on traditional networks cause repeated bottlenecks that constrain network performance.

In a typical data center, some network paths are over-burdened with traffic, while other paths sit idle, resulting in poor application performance and wasted bandwidth. Sometimes separate lanes can be dedicated for special traffic (similar to express lanes for commuters), which is helpful. But for the most part, traditional network architectures lack the visibility and intelligence needed to dynamically optimize application traffic.

Traditional networking’s congestion problem

Traditional data center networking, built primarily for connectivity, is like a two-lane highway that has on/off ramps at major intersections. That network gets you from point A to point B as long as you have plenty of time, and there aren’t too many cars on the road. As the amount of networking traffic increases, more roads might be added, but you’ll still experience traffic jams when too many vehicles try to use the same roads at the same time.

This is the challenge of traditional network configurations that focus on delivering connectivity without broad awareness of the infrastructure. Typically, network switches are used in each rack of servers and storage to provide connectivity for that rack and between racks.

However, these switches only see the data traffic without a broader view of the applications that are communicating, so everything gets thrown together. When enough applications are running at the same time, the result is a congested, slow network – the exact opposite of what is needed in today’s fast-paced data center.

But, what if you introduced a navigation system that could allocate special roads for your most critical application and workload traffic? What if this system could also work to reroute critical applications around congested areas? What if your data highway could be transformed into a single, intelligent, unified network of pathways, allocated not just on connectivity needs, but on the quality of the connectivity needed for specific purposes?

Software-defined automation moves data efficiently 

Workload-aware, software-defined networking (SDN) allows you to do just that. Instead of treating all traffic the same and waiting for highway pileups, the software can intelligently leverage the available connectivity for specific purposes, ensuring the most business-critical applications get access to the least congested roads.

Intelligence and application awareness are centralized in this type of SDN fabric, where a master controller maintains a view of everything connecting to the network. Switching still occurs at the device (switch) level. Yet, with a global view of the roadways and the users of the roadways, the master controller can tell the switches how to better route data traffic to meet desired business objectives.

SDN fabric is like a superhighway system with a myriad of on/off ramps, all controlled by intelligence that ensures the system of roads are leveraged effectively to satisfy the most important traffic. Workload-aware networking software integrates with the rest of the software-defined data center to understand the users of the network and provides operators with the tools to carve out the right resources for business-critical traffic. Additionally, administrators can set up automated processes to accommodate high volume traffic patterns and further streamline operations.

Enter the networking superhighway and leave congestion behind

An intelligently engineered SDN solution, such as HPE Composable Fabric, automatically detects and configures new components as they are introduced to the environment. All components within the SDN fabric are interconnected, allowing any given module to connect to the rest of the fabric with minimal hops. Your data gets on or off the superhighway at a specific destination, without back-tracking or waiting for the next exit.

The efficiency built into this type of SDN solution brings massive improvements to data center performance and agility. Tight integration with the IT environment provides visibility, control, and automation of workloads across the infrastructure, simplifying management and reducing operational silos. All workloads – from legacy to next-gen, scale-out applications – are deployed, monitored, and managed on a single fabric for optimized, evenly distributed resource usage.

Don’t let congestion caused by bottlenecks and boot storms slow down your data center. SDN can transform your network into an intelligent, automated superhighway, empowering developers and speeding innovation.  

To learn more about HPE Composable Fabric, click here. To read IDC’s report about SDN, download IDC Innovators: Datacenter Software-Defined Networking, 2018


About Mat Mathews

matmatthews cropped
Mat Mathews is the Vice President and General Manager for Composable Fabric and Microsoft Azure Stack at Hewlett Packard Enterprise. Mat has spent 20 years in the networking industry observing, experimenting, and ultimately honing his technology vision. Prior to his current role at HPE, Mat was co-founder and VP Product Management at Plexxi. He began his career as a software engineer and holds a Bachelor's of Science in Computer Systems Engineering from the University of Massachusetts at Amherst. To read more articles from Mat, visit the HPE Shifting to Software-Defined blogsite.

Copyright © 2020 IDG Communications, Inc.