Server Room

It’s Too Late For SDN As We Know It

Server RoomAt Silver Peak we frequently have the opportunity to meet with customers to learn about the challenges they face and the initiatives they have. As part of this learning process we ask questions that help us to learn more about the applications they use, how their network is structured, and what specific problems they are looking to solve. Many of our customers use this time with us to also ask questions, learn about how we fit into their initiatives, and what other customers are doing to solve similar problems. One common question that our customers have is: what are you doing for Software Defined Networking (SDN)?

There are many ideas around what SDN really is, but most of them can really be reduced down to a few simple concepts: Simplify and centralize management of the network while providing better control and agility of traffic and resources. Looking back on these meetings, when customers ask us what we are doing with SDN, our initial response to them is very often to ask them the same question, i.e. what are they doing around SDN? We ask because what we are quickly learning is that SDN means many different things to different people, and unsurprisingly, we are getting back a pretty wide variety of responses. Nobody really seems to know what SDN can do for them today, they only have visions about what it might do for them in the future; for most, SDN has yet to translate into any tangible or deployed product.

I believe there is, however, another question we should be asking: is it too late for SDN as we know it?

Before I address this question, let’s take a step back. The epiphany I had (and title of this article) is based on a second question we have been getting a lot: what products does Silver Peak have to address problems in the cloud? As many of us know by now, “cloud” is another term that marketing organizations have been throwing around for a while. However, there is one huge difference between cloud and SDN: while SDN appears to still be in its marketing phase, deployments to the cloud are accelerating at a rapid pace. Through our engagements with customers we are quickly learning that organizations have started making the move from on-premise services into major cloud providers, with some even choosing to forego managing local services altogether for SaaS offerings such as Office 365, SalesForce, or Google Apps. This increasing momentum towards the adoption of the cloud looks a lot like the early days of virtualization, when some non-critical services were first slowly migrated to a virtual platform, and then, as trust was built that the technology worked, more and more mission critical applications were moved. The adoption of cloud services and infrastructure looks to be following the exact same trend.

So why is it too late for SDN as we know it? If the adoption of cloud services continues to accelerate, which by all signs it will — Gartner forecasts worldwide spending on public IT cloud services to be $107 billion by 2017 — what problems are there going to be for SDN to fix in your traditional enterprise network if most, if not all services have been moved to the cloud? In fact, the LAN as we know it today may become nothing more than some basic firewalls, routers and switches that allow users to get to the internet where all of their applications and services are hosted by someone else. If you want to see what many environments could look like in the next five years look no further than to networking giant Cisco and their acquisition of Meraki in 2012, offering cloud managed wireless access points, switches, and security devices that don’t require a CCIE to setup and configure.

Thankfully, there might be hope for SDN after all.

As more of these services are migrated to the cloud, they are often moved much further away from users, and are accessed across internet links which are prone to quality, security, and performance issues. While SDN and its role on the LAN today might be questionable, there is no question that many of the technologies being developed around SDN have greater potential and arguably better application for the WAN than they do for the traditional enterprise LAN.

The move to the cloud is creating new problems that SDN may be able to solve. CIOs will want to provide higher quality services to SaaS providers that they have little or no control over. When problems occur accessing cloud services such as Office 365, the network is going to need to know how to best reroute users based off more than just reachability. Uptime no longer means just being able to get to a service, it means being able to get there quickly and reliably. Security is another growing concern; with data traveling across internet links for all to see, how do you know that your data is getting there safely? Many of the benefits of SDN such as management simplicity, network intelligence, and better control can help to address a wide variety of these issues.

While there are many unanswered questions about how SDN and the adoption of the cloud will evolve, there is no doubt that the new problems, the ones created by moving your services to the cloud, are going to require even greater intelligence, control, and automation of the enterprise WAN. The next phase of SDN — the one that you are more likely to deploy — will need to provide control and visibility to the things that you cannot see or control.

Adam is the manager of System Engineering for the Western US at Silver Peak. He has more than 8 years of experience working with customers and partners on server, storage, virtualization and networking solutions. Prior to Silver Peak, Adam served in System Engineering positions at Emulex and NEC. He grew up in Pittsburgh, PA and is a graduate of Waynesburg University where he earned a B.S. in Business Information Science with a minor in Business Management.