Every day there seems to be a new announcement about the red-hot software-defined
networking (SDN) market.
Technology giants like VMware, Juniper and Ciscoare spending millions, to incorporate this tech into their product lines. There are also a slew of young, innovative startups trying to grab their share of the pie.
What is all the hoopla about SDN technology why should you care?
Let me begin with an interesting parallel. When telephones were first introduced in the US, each town had a manual telephone switchboard, which was run by a switchboard operator. These operators asked the caller what number they were calling, they rang the receiving party, and then manually connected the calls together. As you can see, this process wasn’t private and there was room for human error. Though most people didn’t see a problem here.
Almon Strowger, on the other hand, saw a huge problem. He didn’t believe the existing model of switching calls was private and was prone to human wrongdoing. Strowger, an undertaker from Topeka, Kansas, believed calls intended for his undertaking business were being directed to a competitor. He thought that was happening because the telephone operator was either the daughter or wife of his competitor.
Strowger decided to move his business from Topeka to Kansas City with the hope that a larger city would make his problem disappear. Unfortunately, within a short amount of time, he was convinced his local Kansas City exchange were intentionally giving his customers a busy signal.
Strowger was so frustrated that he hired electromechanical engineers to find a way to eliminate human involvement in telephone switching and this is how the first automated electromechanical telephone switch was invented. Though he started trying to solve privacy issues and intentional human errors in manual switching, his invention had the added benefits of correcting unintentional human errors, increasing the speed of connections, and greatly reducing the operational costs of running a local switching exchange. As the telephone switching industry matured, new technologies were introduced which solved problems of scale and global connectivity.
We’re in a similar period today in the networking industry. Networking still requires teams of network engineers to manage all the networking devices in an organization.
The majority of Software Defined Networking (SDN) vendors are focused on reducing the amount of manual work areas like
- Device configuration (setting devices up)
- Testing and validation (making sure they work well)
- Configuration management (Adds, moves, changes)
- Maintenance (backup, firmware upgrades)
These areas represent the bulk of the operational costs when running a network. By having a way of automating these areas through software, the networking industry is solving problems like:
- Privacy (network isolation)
- Human error (intended or unintended)
- Speed of deployment
- Scale – human scaling as well as device scaling
- More reliable connections which lead to better global connectivity
These were the same problems that Almon Stronger had encountered in 1888.
While the telephone industry took a very long time to solve all these problems, the pace of innovation in the technology industry is quite high and there are several companies looking to solve these complicated problems.
Within the SDN moniker are three distinct approaches to different markets: Datacenter, nework and cloud.
1. Datacenter
Technologies like OpenFlow and products like Juniper qFabric, BigSwitch’s Big Network Controller, and NEC’s ProgrammableFlow are designed to program and control physical network devices within a datacenter. By having a central controller, all compatible devices can be programmed centrally, which reduces operational costs. Since the controller can be accessed programmatically through APIs, the added benefit is that approved applications can now make complex network changes without requiring the direct involvement of the network engineering team.
2. Wide Area Network
Companies like Google are using OpenFlow to build high-performing, highly-available networks which automatically adapt to the changing conditions of the global networks and subsequently routes traffic across the best path available. This approach has the added benefit of maximizing utilization across their network resources. Because of this, carriers and telcos seem to be very interested in this approach.
3. Cloud
An Infrastructure-as-a-Service cloud (when companies rent computer servers and operating systems and upload their own software), creates several problems for the network. The network is how people will access the software hosted on the IaaS servers. Today's networking technology was never designed for IaaS and so its hard for network connections to grow with IaaS as more people try to use the software stored in the cloud.
The technology developed by companies like Midokura and Nicira are working on this problem.
But SDN is bigger than just letting people user faster, less costly networks. By making the network into a fully programmable entity – turning the network into a software platform – a new wave of innovation will be unleashed.
All sorts of new apps will be developed.
Without visionaries and leaders like Almon Strowger, we would have never had the electromechanical telephone switch, which lowered the costs of telephone service, which in turn increased adoption of telephone usage around the world. We’re in a similar time of change and innovation in networking, and the world is starting to take notice.
Ben Cherian is the Chief Strategy Officer at Midokura, a network virtualization company. Prior to Midokura, his last role was as GM of Emerging Technologies at DreamHost, where he ran the cloud business unit. Prior to that, Ben ran cloud company ServiceCloud.
Please follow SAI: Enterprise on Twitter and Facebook.
Join the conversation about this story »
