The long and winding road to a vendor-neutral edge

Network operators have consistently sought what many refer to as vendor-agnostic solutions.  In theory, the model would allow them to avoid what has been called vendor lock-in – therefore offering the freedom to change suppliers, upgrade solutions, or evolve services and architectures without the business or technology limitations one finds in a single supplier environment.  Sounds liberating, doesn’t it?  Let’s do it!

The emergence of virtualization technologies has been a truly empowering enabler of this vision.  The SDN and NFV promise which hit the scene around a decade ago still has broad appeal – but it seems we have an immensely positive value proposition that is unfortunately just beyond the high hurdles of complexity.  A wildly diverse gathering of independent software vendors (ISVs) targeting individual verticals or applications has been exciting, but no doubt created a complex ecosystem.

The reality of this mountain of complexity landed on the shoulders of network operators – both traditional service providers and large enterprises – who mostly found the return on investment tempting but difficult to reach.  This is likely why many of these early-day ISVs moved forward by building their own turn-key branded SD-WAN and security solutions.  It was a necessary move to show value to their target segment.  These ISV pioneers blazed a trail of the possible – and won a fair amount of business along the way.  Well played.

But what about that vendor-neutral network?  A difficult migration – or transformation in this case – that could perhaps be reduced to two phases:

1. Phase 1 would contain the proof; let’s show that it can be done.  It is typically not smooth or sexy, often is noisy and painful, but in the end, you wind up with confidence and a solid foundation.  The technology works and has value.  I contend that SDN and NFV have achieved this milestone.  It has been a combination of open-source options and ISV offerings (from traditional and not-so-traditional providers) ranging from KVM, Openstack, Kubernetes, DPDK and VMware to 6WIND, Versa, Audiocodes, 128T, and Fortinet, just to scratch the surface.  There has also been a heavy push from Intel to provide off-the-shelf hardware that stands up to the demands of high-performance networking.   The point is that the tools for the transformation are now proven and available.
2. Phase 2 may or may not be more difficult than Phase 1, but it certainly has more moving parts.  It requires substantially more than a lukewarm commitment.  Business cases, operationalizing, scaling, improving efficiencies, service definitions and demand creation, and eventually deployments with attached revenue or significant cost savings… it’s difficult work with plenty of opportunities to stall and crash.  And there have been plenty of stalls, crashes, mergers, acquisitions and dissolutions.  The vendor-neutral network is still a lofty goal; however, it is now within reach.  And the benefits in flexibility and control are tremendous – as promised.  Phase 2 will plow ahead – shoulder to the wheel.

The technology foundation for this next phase has been established with virtual network functions (VNFs) abundantly available and performing at a surprisingly high level.  These VNFs have been optimized for multi-core architectures while hardware providers have been optimizing platform performance along the way.  With an open management and network orchestration system (MANO), the primary focus has been to simplify the operations of a vendor-neutral software-defined network.  Careful steps have been taken to obscure the daunting complexity in favor of a very intuitive user interface.  From Day 0 service and hardware templates to Day 1 deployment models through Day 2 lifecycle management of services, hardware, and VNFs, the mission has been to simplify.

Another critical component of performance optimization is delivered via an OS for edge devices.  For example, a VPP-based hypervisor data plane can remove  bottlenecks and enable the construction of high-performance, multi VNF, multi-vendor service chains with a “point and click” process.   The OS is VNF vendor agnostic and further performs quite well on x86 and ARM architectures. The edge OS must be deployable on a variety of white boxes from NanoPi-type ARM architectures to more scalable edge compute deployments.

Illustration of a vendor neutral edge

Illustration of a vendor neutral edge
Source: Telco Systems

So maybe the “long and winding road” has actually led us to the correct destination.  The scenery was different than we expected and there were some unexpected detours.  But the original promise is now ready to be delivered.  It is time to welcome the change.

About the author

Keith Broach is NFV Solutions Director for Telco Systems. Broach started his networking career with Southwestern Bell Telephone back when “voice was king.”  He has participated in several key infrastructure transitions in the networking space, most recently in the service orchestration and virtualization domains.

Keith has held product, engineering, operations, and sales roles in fiber optic transport, IP networking, virtualization, and management and orchestration fields at Cisco, Cyan (Ciena), 6WIND, Cobo and Telco Systems.  He holds a degree in Mechanical Engineering from the University of Arkansas.

DISCLAIMER: Guest posts are submitted content. The views expressed in this post are that of the author, and don’t necessarily reflect the views of Edge Industry Review (EdgeIR.com).

Article Topics

edge hardware  |  MANO  |  network management  |  networking  |  NFV  |  SD-WAN  |  security  |  Telco Systems