Arthur Cole spoke with Dante Malagrinò, president/CEO, Embrane Inc.
Most discussions surrounding software-defined networking (SDN) make it seem rather monolithic — like there is a single way to architect and implement, and it will deliver largely the same results across all enterprises. But as Embrane’s Dante Malagrinò points out, there are many flavors of SDN out there. Embrane’s, for example, functions on layers 3-7, which puts it closer to the application and is therefore less disruptive to existing infrastructure. The goal of SDN, he says, should be to improve legacy systems, not replace them.
Cole: Software-defined networking offers great promise, but little is said of the possible consequences of such a fundamental change in network architecture. What are some of the key concerns the enterprise should keep in mind as they assess their SDN strategies?
Malagrinò: Everyone talks about the agility and the savings benefits of SDN. However there is one area that we find to be most often overlooked by early adopters, and by not paying attention to it, can cause some significant disruption to the business. It’s around making the adoption of SDN technologies easier to digest. SDN should not be about working around the network team, but rather giving the network team the tools they need to be as agile as their compute and storage counterparts. It’s important that SDN solutions be aligned with existing operational models – organizational structures, tools and processes. Taking a non-network-centric approach to building networks will ultimately cost a company significantly more than the implied savings.
Evaluations, purchasing, deployment, maintenance and trouble-shooting of SDN technologies should be conducted by existing people, tools and processes, which are less disruptive and costly than performing these functions by other organizations. The networking team has accumulated vast experiences that should be leveraged, not worked around. If/when the network breaks, do you want the systems teams to fix it? Would you want your network team to fix your applications when there is a software bug in an application?
The other concern is more about something enterprises should clearly understand, and it’s that they don’t need to do a wholesale change and virtually rip and replace everything at once. Companies have made investments already and they should look for solutions that can work with those existing technologies. In fact, they can dip into SDN by doing something like augmenting layer 3-7 network services, such as firewalls or load balancers without any disruption whatsoever. They will still get the agility and cost advantage without drastically altering their network or operational environment. Enterprises can also start leveraging SDN in specific parts of the IT department. One area is the development/test environment where, again, adding software-defined network services that mirror the production environment bring advantages without disruption.
Cole: Embrane's heleos system takes a somewhat different approach to SDN. What did you set out to accomplish?
Malagrinò: We’ve taken a top-down network-centric approach to SDN vs. a bottoms-up look that often focuses on instilling networking technology in the hypervisor. In other words, we think there is greater value in developing software-defined network services that sit closer to applications and keep the network functionality within the network. In terms of the “network stack,” Embrane focuses on layer 3-7 network services instead of basic layer 2 connectivity, which seems to be a more common definition of SDN.
Embrane can augment or replace existing hardware firewall and load balancer appliances non-disruptively while delivering the advantage of agility and cost savings that SDN promises. Embrane infuses agility into the network by enabling IT to spin up and non-disruptively grow or shrink network services such as load balancers or firewalls in seconds – something neither hardware appliances nor virtual appliances can do. Embrane also supports operational simplicity by augmenting existing operational models and best practices, giving a single pane of glass for easier management of multiple network services and offering the tools to build high levels of automation and self-service into the network and as part of public/private/hybrid cloud architectures. From a pure savings standpoint, Embrane saves companies at least 45 percent in CAPEX versus hardware appliances, not to mention the operational savings they will achieve. And finally, our elasticity feature also eliminates unnecessary overprovisioning and simplifies capacity planning, which again is impossible with hardware appliances.
Cole: Ultimately, though, we're talking about a highly dynamic network environment that can be configured and reconfigured on the fly. Won't that require a fair amount of disruption to existing networks no matter how it is implemented?
Malagrinò: At the end of the day, all SDN players are trying to solve the network agility problem. If you speak with customers though, they want ways to rapidly provision and configure the network and they want to build levels of automation and self-service. Introducing anything new requires change. The trick is to minimize it where you can. We have taken the approach of making it less disruptive to adopt, support and integrate our software-defined network services. Aligning with existing operational models and organizational structures is one way to reduce the pain of disruption, and deploying technologies that integrate and/or interoperate with your existing infrastructure investments will also help reduce the disruption and cost of making significant changes and improvements to the network.