Posted by Mae Kowalke on Tuesday, May 19, 2015 with No comments
Early adopters of software-driven technologies have reaped significant efficiencies and economies of scale. Software Defined Networking (SDN) is just one piece of the larger Software Defined Everything (SDx) evolution, which will transform all IT infrastructure, SDxCentral predicted in the first part of a series on the topic.
Much more than just networking is being transformed by software, after all; this change is affecting physical items and functions, and there are value-shift implications beyond the technology industry.
The change is causing infrastructure companies to undergo an identity crisis, SDxCentral said. They are asking themselves big questions like, Who are we? What and where do we deliver value? The outcome of answers to these is murky, although some predictions can be made based on current discussions.
In the telecom network world, SDN is ultimately all about apps, said Dan Pitt, Executive Director at Open Networking Foundation, in a Light Reading article.
What he means by that is: operators use SDN apps (software programs designed to perform specific tasks in an SDN environment) to keep up with user expectations by controlling QoS and managing increased traffic.
Pitt observed that, as with consumer app stores for mobile devices, network developers now can create SDN apps to help service providers cut their OpEx and CapEx and deliver new monetization opportunities.
Indeed, the initial technology hype staged has passed for SDN (at least in the carrier network world), and operators are now grappling with implementation and commercialization questions, Light Reading noted.
Some of the big topics on the minds of service providers right now, Light Reading said, include real-world SDN use cases, IP and optical integration, SDN for cable MSOs, and SDN security.
Relatedly, SDN and its companion NFV are creating on-demand bandwidth capabilities with transformative effects for providers to offer flexible Ethernet connections to both public and private clouds, as Level 3 is doing, Light Reading noted.
Level 3's goal is to solve customer challenges hybrid cloud, such as extending dynamic bandwidth control to cloud connections. They believe there is a solid use case for enterprises traversing between private and public clouds for at least some of their utility computing and storage.
The key here is how enterprises can efficiently and cost effectively move applications and workloads from on-premises data centers to multiple clouds. They want to do so in a way that doesn't require them to scale compute/storage and network capacities separately, Light Reading stressed. This is a big opportunity for providers, and those like Level 3 are on the leading edge of a new service market.
Meanwhile, there's an SDN-related revolution going on in data centers. SDxCentral explained that for some time network vendors have been focused on keeping up with exponential traffic growth and changing service needs. But, revolutionary changes are still needed for network programmability, policy-based provisioning, and automation. SDN can fill that gap--if implemented with care.
One side effect of the hype over virtualization is that hardware tends to get downgraded to second-class citizen in data centers, SDxCentral noted. Hardware is becoming commoditized, and software wants to be hardware independent. But, its dangerous to take this too far; hardware is still needed.
Best practice lessons from the past suggest that efficient virtualized data centers will depend on software and hardware working together, SDxCentral stressed.
It's unclear yet what abstraction techniques will work best to allow hardware innovations to shine. Single Root I/O Virtualization (SR-IOV) and Open Virtual Switch (OVS) are two possibilities, with trade-offs between them as with so many other areas of IT where the technology selected depends on which compromises are acceptable.
Where do you think SDN's biggest impact will be in the next several years?