To say that software defined networks has been hot is as gross an understatement as saying Boston sports team are better than Chicago sports teams. The statement is so obvious, you're almost embarrassed to say it. However, while nothing's changed on the sports front (Theo, enjoying the Red Sox run?), I do believe things are changing in the world of SDNs. Initially, the market focused on the economics of networking and programmability, but it seems recently the market has shifted the focus of SDNs to network function virtualization (NFV).
For those not familiar with the concept of NFV, this technology allows for specific network services to be virtualized and then run on a hypervisor. The concept actually isn't new. When I became an analyst way back in 2001, my first report was on a category of products called "IP Service Switches" that enabled things like network-based firewalls, routers, etc.
This market crashed and burned rapidly. Why? Well one of the big issues is that these products were very expensive--some starting at a few hundred thousand. Virtualization technology was something that geeks played with and was in no way mainstream. Today, the entire economic model of the server industry has been turned upside down, as virtualization technology is now robust enough to handle even the most demanding workloads, enabling cost-effective NFV.
One vendor taking advantage of the advancements in virtualization as well as the market opportunity for NFV is Sonus. On Wednesday, Sonus announced the Sonus SBC SWe (Software edition). The name may be highly unimaginative, but it is accurate, as the SWe delivers a fully-featured SBC on a virtualized platform.
When it comes to SBCs, I do think there is a real need for a lower cost, virtual version of an SBC. Traditional hardware-based SBCs are relatively expensive, and many large enterprises pick and choose where to deploy them. Much of the mid-market and small market just can't afford it, so a virtual version will broaden the appeal of SBCs down market, but also widen the deployment opportunities within large enterprises.
In fact, what I've seen in other markets that have shifted to virtual appliance (application delivery controllers, WAN optimizers, etc) is that the virtualization and on demand nature of the appliance opens the market up to application developers, DevOps and other groups that may not have necessarily bought that appliance before.
As I said, there is definitely need for software-based SBCs and I do like a number of things about the Sonus SWe including:
* Common code base with the hardware appliance. It's not uncommon to find a hardware vendor that releases a "virtual edition" of a product with a very limited feature set and a totally different operating system. Some vendors have many hardware dependencies and can't create feature parity. Others may not want to cannibalize the traditional hardware devices, so the feature set of the virtual appliance is purposely "trimmed" to incent customers to continue to buy the hardware version. In this case, the Sonus hardware and software versions have identical features and code base, meaning customers don't have to make choices based on features rather on form factor.
* Ability to start small and scale up. The low end of the SWe offering is 25 sessions. Assuming a 10-1 rate of users/session, the product scales down to businesses as low as 250 employees. Once customers have deployed the product, Sonus provides an easy way to scale the product up to unlimited number of sessions.
* Hardware and VM independent. This is the real test of a virtual appliance vendor. If it's truly a virtual appliance, then the vendor shouldn't really care what hardware platform or VM the product runs on. I believe SWe is completely hardware agnostic and has been tested on all of the major hypervisors.
The release of the software-based product does raise the question about what Sonus does with its 5000 series hardware appliance? Is it dead? Does it go away?
In fact, I'm not sure why the industry likes to fixate on one or the other; there is plenty of opportunity for both. The decision of whether to use the hardware version or the software version is based on what's important. If absolute, predictable, dedicated performance is the most important attribute of the SBC, then pay up and buy a hardware appliance. Large enterprises, service providers, UCaaS providers and the like need the dedicated performance and will lean towards the hardware products.
However, if agility, flexibility and on-demand provisioning are more important, then buy the software version. This would be true for most mainstream enterprises that don't use near the capacity of the current hardware products. When it comes to communications, performance = hardware, while agility = software. I agree there's lots of overlap between the two, but that's a good general rule of thumb.
I fully expect the "software-ization" of the SBC to act as a catalyst to get the product deployed in more places in more companies. Sonus has been trying to become a more enterprise-relevant company over the past few years, and the SWe should be a big help in getting to that goal.
Follow Zeus Kerravala on Twitter and Google+!
@zkerravala
Zeus Kerravala on Google+
