To SDN Or Not To SDN -- There Is No Question

Regardless of all the news and hubbub around Software Defined Networking (SDN) today, adoption is a little slow.  Not stagnant, but just a little slow.  This is understandable.  SDN is new and is a pretty significant departure from traditional methods of designing, configuring, deploying, and managing networks and network functions.

But of course it is a significant departure; it is virtual networking after all.

This tepid acceptance of SDN is not so different from the initial lackluster acceptance of server virtualization just over a decade ago (x86 virtualization that is); and yet today not virtualizing servers is unimaginable in almost all cases.

(I would argue that today there are still workloads on physical servers, for various reasons, that need not be--showing this tepidity can last 10s of years or more)

However, SDN is quite probably even more innovative, more revolutionary, and more disruptive than server virtualization ever was.

Once companies and organizations begin to dip their toes into the SDN waters they will realize a whole new plane of existence within the data center has opened up to them.  Gone will be the days of having to wait days, or weeks, for new network contexts to be provisioned.  Gone will be the days of having to acquire [as much] physical networking infrastructure for simple network changes or updates.  Gone will be the days of having to worry about complex network upgrades and the risk they pose to the data center.

More than that--more than what is gone when using SDN--is what is available.  When using SDN, virtually (pun intended) any network configuration is available to the customer, network designer, data center administrator, everyone.  Within a few clicks an entirely new virtual network can be created and made available.  This virtual network can be modified, updated, and changed with just a few more clicks.  Multiple virtual networks can be made up of the same network space, without conflict, even within the same virtual environment.  Virtual servers can be moved amongst the networks in a matter of seconds.  Network Function Virtualization (NFV) brings about virtual DHCP, load balancing, firewalling, and more.  All equally as easily modified with the click of a mouse.

Imagine what this means.

Imagine an application and all its VMs, networks, firewalls, and configurations being cloned and restarted elsewhere, with no conflicts.  These new Virtual Data Centers (VDC) can be used to test upgrades, new code, configuration changes, or security compliance scanning without affecting the production workloads.  Forensics can be conducted on these VDCs in the event that there is a concern regarding viral infection, hacking, or other malicious activities, and because the VMs and their network/security configurations are exactly the same the forensic evaluations will be perfect.

Imagine cutover of a workload to a new configuration, application update, or upgrade with a very, very, short "maintenance window."  Not a window of hours, or days, but of milliseconds.  Because the entire VDC is cloned and has all updates and tests performed on it without changing anything or affecting the production VDC it is possible to simply moving a virtual route from the current production VDC to the updated one again, in a matter of clicks.

Failback in the event of a problem, just as easy.  Just move the virtual route back.

Also imagine networks which are more than just highly available.  SDN-based virtual networks scale with the VDC environment.  These virtual networks can have tens, or hundreds, of nodes providing service, instead of just a couple of clustered physical networking devices.  Forget high availability; this is massive availability.

Unfortunately, there appears to be an impression that SDN is something designed for, and only manageable by, extremely large data centers of the likes of a Google or eBay/PayPal.  This couldn't be further from the truth.  SDN is actually quite inexpensive, when compared to physical infrastructure and its support contracts, and is easily managed by those willing to adopt it.  It can benefit almost any company or organization of almost any size.  

These examples are but the tip of the ice berg:

• Extreme reduction in time required to update or modify network configurations
• Reduction in cost of required physical infrastructure
• Reduction of risk during upgrades or updates of applications
• Increase in security and forensic capabilities
• Massive availability
• Benefits companies and organizations of all sizes, small to large

Beyond this, the sky is the limit.  It isn't possible this early in the game to predict exactly what can/will be done when using SDN.  Its implications on the WAN and with regards to endpoints is thus far unknown; but it is developing, and it will be amazing.

At the end of the day, the company or organization that does not choose to adopt SDN is at a disadvantage.  They will find that they are more costly and less agile than competitors who do adopt SDN.  This is not just applicable to companies providing IT services either.  Regardless of industry or size, every company and organization can benefit from SDN by way of lowering infrastructure costs and increasing their ability to rapidly adapt their network, and associated business applications, to meeting their customer's ever evolving needs.

In closing, it is not a question of whether or not to adopt SDN; it is a foregone conclusion.  The question is only a matter of whether or not a company or organization chooses to adopt today, and begin reaping the experience and benefits, or if they will adopt later (if they are still around), after years of continued expense and risk due to the utilization of traditional physical networking infrastructures.