Is SDN Flawed?
Many have grown weary of the term “Software-Defined Networking.” It seems we’ve discussed the concept for many years, but very few organizations have taken SDN past contemplation and into production. Why is that? Is SDN somehow flawed?
In a recent discussion with Jim Metzler, CEO of Ashton, Metzler, & Associates, I asked him that very question. Jim recently wrote a paper on NFV/SDN for TM Forum, so he has been engaged with many of the key players in the SDN arena (you can download the TM Forum paper for free, compliments of SevOne).
Jim put the issue of SDN adoption into perspective with this comparison:
“Let’s look at VMware. An overnight success, right? Well, actually VMware was founded in about 1998. It began to get traction about 10 years later. If I look at OpenFlow as a piece of SDN, Martin Casado got playing with that back in 2006-2007. So I can make the case that we’re only in year eight or nine of SDN, and we are getting very close to crossing the chasm.”
Still, the adoption of SDN is not without concerns:
- SDN is complex – while the abstraction of the control plane from the forwarding plane sounds like a logical evolution, the reality is that SDN is inherently complex. Understanding dependencies and constraints in dynamic, virtualized environments is a supreme challenge.
- Benefits difficult to quantify – not too long ago, it was thought that SDN may bring significant CAPEX savings. That belief has since taken a back seat to the primary driver of SDN adoption: agility. However, agility is difficult to quantify and measure. Proving its value can be elusive.
- Cultural change – many have argued that the greatest challenge posed by SDN is not technical; it’s cultural. It represents a fundamental organizational shift in the roles and skills required to make SDN work. For IT staff groomed on command line interface configurations, this presents a challenge.
Perhaps the flaw is not with SDN, but with our own expectations. Why is it that, with a concept so inherently complex, we feel it should have already “crossed the chasm” into mainstream adoption in such a short timeframe?
The truth is, we’re close. An article run by Network World back in September, 2014 suggests that we may have already passed the tipping point with AT&T’s introduction of “the first software-defined networking solution of its kind in the United States,” scheduled to deploy this year. According to the article, AT&T’s move represents a “paradigm shift that enables business users to self-provision services from communications services provider using online portals.”
As SDN and its partner NFV near mainstream adoption, networking teams will place further emphasis on the need for monitoring the performance of these new solutions. There isn’t a tested roadmap for successful deployment. It will require close scrutiny at every phase to ensure proper service delivery and a return on the investment. In the future, we see performance monitoring platforms as an integral part of the SDN ecosystem, providing the controllers with insight as to how to optimize service delivery.
So SDN is not flawed, but it does represent significant change. And with change comes risk. Do not underestimate the role performance monitoring plays in mitigating that risk.
Interested in learning more? Download our whitepaper on Top 3 Challenges of Monitoring Software-Defined Everything.