As part of my job as an analyst, I talk to a lot of IT, networking, and security professionals. Based upon these conversations, I have no doubt that data center networking is going through a profound transition. Why? Simply stated, the old model of static inflexible physical networks can't keep up with new requirements like rapid provisioning, server virtualization mobility, low-latency, and massive scale.
Unfortunately, while networking professionals realize that data center networking changes are imminent, they are extremely confused about what's coming next. Flat Layer 2 networks? Data center fabrics? Distributed virtual switching? Data and storage network convergence? VEPA? OpenFlow? TRILL versus SPB? I live in this world and find that its hard to keep track of all the moving parts. Pity the poor network architect or engineer who has to make sense of all this stuff while still performing his or her real job: Keeping the packets flowing.
I wanted to sort through the data center morass myself, so I ended up writing an ESG Market Landscape Report on the topic. The goal of the report is to guide users through the evolution of data center networking, define some of the impending standards and what they mean, and provide a basic roadmap on where things are going over the next few years. The report is available for free at this link:
To be clear, this report is not a Gartner MQ-style publication that rates vendors. Rather it is meant to help users better understand the technology and vendor landscape. Special thanks to Arista Networks, Avaya, Brocade, Cisco Systems, Enterasys Networks, Extreme Networks, Force 10 Networks (now Dell), Gnodal, HP, IBM, Juniper Networks, and NEC for providing a ton of background information. These leading networking vendors really gave me a rapid education on the topics covered in the report.
Finally, data center networking is a topic that is both broad and deep. I know there are things that I missed and there is always so much more to learn. For example, after publishing the report, I had a great dialogue with Arista about Layer 2 versus Layer 3 network considerations with clustered file system applications (i.e. Hadoop, MapReduce, etc.). I welcome all feedback and comments either here in my blog or via email at email@example.com.