The rise of technology trends such as big data and BYOD (bring your own device) are causing a seismic shift in the way data centers are managed today owing to the massive amounts of data they drive. To handle this increasing amount of data, IT managers are turning to software-defined networking (SDN), a new approach to designing, building and managing virtualized networks that separates the network’s control from the physical hardware. SDN is viewed as a manageable, cost-effective and adaptable architecture that can handle the high-bandwidth and dynamic nature of today’s applications.
At the most basic level, SDN is a networking architecture that serves as a software-based “traffic cop” for all the information that comes through a data center. The software, also known as the SDN controller, provides a central view of the overall network and enables IT managers to control and configure routers and switches without having to manually configure each physical appliance. With a central view, IT managers have a clear picture of their entire network, which in turn makes management easier and provides more control over network traffic.
But although SDN delivers the benefit of an agile, manageable network by optimizing performance, many organizations may not realize the significant impact that it can have on the hardware in a data center. The more businesses virtualize their applications, networks and infrastructure—depending more on remote software and less on hardware—the more critical data center uptime and availability become.
The Impact of Software and Virtualization on the Physical Data Infrastructure
In a software-defined system, where network, storage and servers are virtualized, the defining points of the environment are no longer the physical devices. By shedding physical constraints, these traditional IT assets become more dynamic and flexible—but also more critical. As a result, IT personnel have begun to shift their focus to managing the increasingly important IT assets: applications and software. Because of this trend, the data center manager has also shifted focus, taking into account the variables that IT infrastructure needs (the rack space/availability, power and cooling) to ensure they continue to run smoothly.
Virtualization, cloud computing and software-defined networking can have a major impact on the reliability of the physical data center infrastructure. In a software-defined data center environment, traditional systems are seldom agile enough to power and cool elastic virtual switches, for the following reasons:
- High densities and hot spots can arise: Virtualization often leads to higher power densities. Even though virtualization may help reduce overall power consumption, virtualized equipment is installed and grouped in ways that can create local high-density areas that can then lead to hot spots. If not addressed, these challenges can threaten the reliability and availability of the data center.
- Rack-level power and cooling must be considered: Virtualized IT loads—particularly in highly virtualized cloud data centers—can vary in both time and location. So to ensure availability in these systems, rack-level power and cooling health must be considered before changes are made.
- Virtualization negatively affects PUE: Virtualization reduces IT loads, meaning the data center’s power usage effectiveness (PUE) is likely to worsen. Right-sizing both the power and cooling infrastructure to the reduced load can help improve PUE.
Elastic, versatile equipment is needed to efficiently cool and power the flexible, dynamic nature of software-defined networking. By using modular, high-performance power and cooling systems, data center managers can help ensure optimal uptime and availability while avoiding unnecessary downtime.
Converged Infrastructure: A New Approach to IT Infrastructure for SDN
Cooling and power systems are not the only pieces of physical infrastructure that can use an upgrade when it comes to managing software-defined networking. In fact, in a post-virtualization software-defined data center, old legacy IT stacks aren’t cut out to support the needs of virtual workloads. Since siloed physical storage and network assets lack the optimization to support virtual servers, resource overprovisioning may result. Throwing more hardware at the problem only adds more complexity and cost, and it doesn’t fix the issue at hand. Therefore, a new approach to IT infrastructure is needed.
Consolidating IT infrastructure components into a single optimized platform with central management—a system commonly referred to as “converged infrastructure”—can enable increased utilization and lower costs. In this way, benefits that come from software-defined networking can be realized.
Where legacy IT infrastructure often falls short, converged infrastructure allows you to design, build and maintain segments of the virtualization stack while supporting growth.
The Importance of Data Center Infrastructure Management in the Software-Defined Era
In addition to implementing a converged-infrastructure system, data center managers should consider integrating data center infrastructure management (DCIM) software. DCIM provides single-pane-of-glass visibility across the entire data center and is therefore able to inform the SDN software where to send packets for optimal efficiency and availability. Without DCIM, SDN creates a more complex and critical environment that can be difficult to manage properly, putting data centers at risk for downtime and stranded capacity.
The software-defined era is turning business models upside down, shaking up the way businesses run and computers compute. And every new technology implementation comes with new complexities and challenges. Businesses will need to approach their infrastructure and management in a new way to take advantage of SDN. Therefore, data center managers must ensure their facilities remain agile and flexible to meet the current and future needs of businesses.
Leading article image courtesy of cbowns
About the Author
Himanshu Patel is a senior manager for APC by Schneider Electric, a global specialist in energy management. Himanshu is responsible for developing channel strategies and developing relationships with strategic alliance partners. Before joining APC by Schneider Electric, he worked in various roles including sales, channel management, channel marketing and consulting for a range of organizations from Fortune 500 companies to startups. Himanshu holds a BA in political science from the University of Massachusetts at Amherst.