Who is going to win the SDN battle?


On April 28th 2015, Interop named as Best of Interop 2015 SDN Winner, "Cisco – Application Policy Infrastructure Controller (APIC) for Cisco Application Centric Infrastructure". 
For more info, here is a video presentation of Cisco APIC by their director of product management.

The three SDN finalists were:
This award comes 9 months after Cisco released its ACI offering, in the past year, Cisco seems to have evolved in the Software Defined Networking (SDN) arena.
Back in 2014, VMware was the only vendor capable of delivering a solid SDN product that customers could use immediately, and it was shipping this product to customers around the world.
Cisco's APIC offering started shipping in August 2014 with an intelligence in the architecture spread between the controller and the infrastructure devices (physically distributed policies using the OpFlex protocol), an architecture different from its software competitors which have centralized the intelligence in the SDN controller (Like OpenFlow based solutions or VMware NSX).
Soni JiandaniSVP Marketing, declared at Cisco Partner Summit 2015 (#CiscoPS15) in Montreal last week, that Cisco has sold ACI to hundreds of customers worldwide (the largest being located in the middle-east), that's not bad for a product that has been shipping for less than a year.


One of the key reason why ACI is gaining momentum since it was released, is that Cisco has managed to partner with its own competitors (like F5A10, CitrixFortinet, Checkpoint or others), in order to deliver an open system and protect customers' investments in these technologies. Customers opting for ACI will be able to automate their mission critical application workloads across all those platforms, therefore maintaining the features developed on all those layers.

It seems today that Cisco is delivering on its promises which makes us wonder how things will unfold with its direct competitor, VMware. I believe that for the core enterprise and service provider networks, the story of Cisco will prevail since it will be able to capitalize on its existing deployed base and the need for wire-speed forwarding and high-performance networking at 10G, 40G or 100G. The adoption of ACI in the existing Cisco networks will also be driven by a need for a tighter security and more visibility on the hardware infrastructure. This will not eliminate the possibility to run NSX over ACI as an application, a possibility which is still available to anybody who would like to run it.

VMware has also developed some technology partnerships with PaloAltoCitrix or F5 around NSX. Although the possibility of applying the security policy directly to the virtual machine (VM) is quite interesting, it is still unclear how the system will be able to take wire speed 1G/10G/40G east to west traffic between the VMs without controlling the actual hardware.
Smaller non-Cisco networks with less performance and network visibility requirements, would find more value in  VMware NSX which has the unique capability to run over generic hardware.

In the 1980s, Alan Kay ,one of the greatest software designers, said : "People who are really serious about software should make their own hardware", Apple is one example where this statement has proven its value, they designed the hardware that works best with their software, hence the reliability and worldwide appeal of their products. Google followed suit by building their own network and computing platforms. It is a strong argument supporting the Cisco ACI strategy.

Cisco ACI provides the interface which would transform the network from configurable to programmable and automated (through its controller), it will not though provide companies with IT orchestration despite the fact that it interfaces with all major automation and orchestration systems (like VMware or IBM for example). Still it accommodates the APIs necessary to interface with all those tools in the market. VMware's strong point is the fact that it delivers control, automation and orchestration while its weakness is that it does not control or monitor the hardware infrastructure.

Cisco evolved and closed the existing gap with VMware, if the later wants to regain its market edge, it needs to develop interfaces with the different hardware technology providers who have been very innovative in the past years, particularly when it comes to security and intelligent routing. Software running on generic hardware will no longer cut it, it is not as straightforward as the x86 virtualisation journey, where limited computing technologies were at stake. It is about advanced networking, complicated security and accurate QoS, the three pillars that either make the outcome exciting or very challenging for every network engineer out there.

At the end of the day, SDN is a multiplayer arena, to be the undisputed leader requires a solution that encompasses the largest technology ecosystem,  coupled with a robust capacity to deliver an end-to-end system.

If you are interested to learn more about this topic, you will find lots of technical information online, via slideshare and by following our future posts on Data Consult, here is a nice article from Raed Hmaidan.  We are still at a point where customers are experimenting with what is best for their environment and we invite everyone to ponder upon this issue.  It is the beginning of a journey where we will see automated applications that will impose their own infrastructure decisions. On which platform, it is unclear yet, however when it comes to deciding on critical applications, any decision needs to be based on full visibility of the underlying software, hardware and end user infrastructure.

Marc Nader
Chief Operations Officer
@mourcous
lb.linkedin.com/in/marcnader
Previous
Next Post »