Cloud & Virtualization Require SDN

With the constant flow of emerging technologies, it's easy for organizations to meet resistance when trying to adopt the "next big thing" in enterprise IT. From big data, to cloud, to software-defined everything, enterprises are bombarded with options that seem to promise development and optimization of key business processes.

Over the last few years, IT teams have been flooded with talk about virtualization and cloud. Virtualization of applications and compute resources are widespread, but they aren't the only considerations in taking advantage of cloud computing. As I outlined in my last article, the cloud represents the ideal consumption model for enterprise information and communications, but there are some concerns including security that need to be addressed.

As businesses begin to invest more in creating the most suitable cloud environment, they need to take a closer look at their entire infrastructure -- especially the network. Not every cloud infrastructure is ideal for mission-critical business applications due to limits in responsiveness, control and visibility. Most large enterprises are geographically distributed and use applications that are time-sensitive, persistent, or media-based. This challenges quality of service when delivered over the network.

A crucial factor in delivering a high-performing cloud environment is virtualizing the network. Simply put, a better network drives a better cloud. In the broadest sense, the network is the connection between servers and storage clusters within and across data centers that ensures the performance and availability of applications and data. It is a fundamental contributor to users' and corporations' quality of experience, without which the true potential of the cloud will be compromised.

Many enterprise CIOs are living with the pain of the network today. While compute management systems assign virtual machines (VMs) within seconds, it can take hours or days to provision the network to deliver optimal paths between new applications and users. There is a mismatch between the operational agility of data center networks and the applications that they serve.

Server virtualization has made computing resources easily consumable, with the ability to turn up virtual machines in minutes and move or release them as required. Servers once dedicated to a single application are now expected to run 40 or more virtual machines. This shift bears a direct impact on the network through the explosion in the number of virtual endpoints, and is compounded by the dynamic nature of virtual machine instances.

Network instantiation today is highly manual and work order driven, with configuration performed device by device in the data center network. The static, highly manual, configuration-driven networking model that has sufficed to date must yield to a new paradigm of dynamic network service auto-instantiation. Such a shift will increase service velocity and result in dramatic improvements in overall operational efficiency.

This is where a new paradigm of networking, software-defined networking (SDN), comes in. With SDN, networks are transformed into an open and programmable component of the larger cloud infrastructure. I previously talked about the emergence of SDN as of the three industry trends transforming enterprise IT.

SDN gives businesses more control of their infrastructure, while reducing overall capital and operational costs. SDN can virtualize the network -- first in the data center where the requirement is most pronounced, and over time extending to the wide area network -- to meet the needs of private, public, and hybrid clouds.

Two key benefits of SDN are the ability to simplify the network by abstracting its capabilities and deploying applications more quickly, as well as the ability to eliminate provisioning errors and increase security and compliance through policy-based automation.

Abstraction through SDN can shift the way applications away from the "lost in translation" process in which apps are forced to deal with the network implementation detail. Instead, we can utilize an IT-friendly design of network services based on a simple IT language for defining the network to accommodate application needs.

Automation involves rethinking how network services are activated. We must move from a configuration-driven approach with associated layers of operational complexity to a policy-based auto-instantiation model where network services are deployed using defined business and network templates.

For cloud services to thrive, the network infrastructure must become as dynamic, as virtualized, and as consumable as the compute infrastructure. Increasingly companies will enjoy the greater business agility and operational simplification of a virtualized network, using SDN, to enable agile cloud environments that are responsive, policy driven and highly automated.