Is Cisco fighting a losing battle over SDN?


Boxing_GlovesI’m wondering if Cisco is fighting a losing battle with their ACI approach to SDN.  Basically, they are looking to provide programmability to the network via their Application-Centric Infrastructure.  This is different from the current software first movement in that Cisco’s approach is based on their hardware first view of the network.  However, if you were to go by the response to the number of vendors supporting the NSX ecosystem you’d think the entire industry is going against this hardware first religion.

All of Cisco’s major vendors are lined up in support of VMware’s NSX software based virtualization solution.  The list includes a who’s who of Cisco competitors.

  • Arista – Top of Rack
  • Juniper – Data Center Core
  • Dell – Closest & Top of Rack
  • HP – Top of Rack
  • F5 – Security/Load Balancing

The above list is just the names I saw as I walked the conference floor.  There was a time when Cisco could make a decision before the rest of the industry or against the general will of the rest of the industry and the industry just had to follow suit.  I don’t believe this is a similar situation.  Cisco has always seemed to be one step ahead of the industry and fulfilling the needs of their customers by creating protocols that didn’t exist or extending existing protocols to enable greater functionality.  However, Cisco has been slow to react to the drivers of for SDN.

The largest of tech companies have embraced some form of SDN, most of which home grown as I understand it.  Along with the march of their competitors toward Software Defined Network or better Software Managed Networks, I see a point where vendors like VMware will just manage around Cisco’s ACI enabled hardware.  I predict environments where customers will have hardware designed to take full advantage of Cisco’s ACI architecture but will just be managed by some other SDN solution that integrates with multiple network hardware solutions.  VMware already produced a great post on how to leverage NSX in a Cisco UCS environment.

I understand why Cisco has decided not to jump aboard the NSX train.  I just wonder does it still have the industry might to continue and buck the rest of the industry.

Published by Keith Townsend

Now I'm @CTOAdvisor

4 thoughts on “Is Cisco fighting a losing battle over SDN?

  1. Really great post Keith!

    Honestly I think the single biggest challenge Cisco has against it is wall street. Their massive margins are no secret. Dell for example makes double Cisco’s revenues and yet makes only half the profit. People have been publishing for years that Cisco makes ~80% margin on many of their core products. This is essentially their innovator’s dilemma. The 6500 for example was a great box and allowed them to continue to position old technology with ridiculous margins – to this day customers still buy first generation 6500 linecards. And maybe for some uses thats ok, but it creates a huge challenge for Cisco in a market that is continuing to increase in competition. With wall street the way it is, every big company has to turn in double digit growth every quarter or get massacred by wall street. That is a much bigger problem than Cisco. But today for Cisco customers, this is a really important issue – customers want and need better deals. Cloud providers are now co-op-iters with IT departments and those cloud providers are using cheaper, much more efficient and much better performing devices – IT departments simply cant afford to use substantially more expensive & less efficient gear than cloud providers if they dont want the infrastructure outsourced – which is a fine model also. I think the center of the matter is that, its pretty clear cloud providers are deploying the most sophisticated data centers with the most sophisticated network use cases. It could easily be said that most of what enterprises want to do in their data centers are all things that cloud providers have already done. So it should cause alarm that cloud providers are already doing much higher density, highly automated elastic cloud networking, and they are doing it all on low cost boxes with no proprietary tech that also uses substantially less rackspace, power, heat output, better performance etc. If next-gen networking is already being done on these devices, why would Cisco position an alternative approach that is fundamentally worse by all measures of performance? That is where Cisco’s innovator’s dilemma comes in. The writing on the wall is that networking is poised to move towards an x86 like hardware model. The bottom line is that Cisco knows they could not maintain their margins if this became popular – so they are in a fight against it. That is my beef against Cisco, they have great technologists and could do great things, but every technology they put out reeks of putting their own business needs ahead of customer needs and ahead of the best they could be doing with technology – and they think they are enough of a marketing powerhouse to pull that off. I hope that doesnt work, and much more than that though, I wish wall street would get fixed so every company didnt have to be in this predicament, its unhealthy for every industry, horrible for society, and frankly I think its just sad for Cisco. At the end of the day, the market needs a reset with many vendors each fighting over diverse market shares, so no company ends up in the position Cisco is right now.

    1. I’ve wondered out loud if Cisco has just been positioning itself for the new reality of lower profit margins from SDN in general with the layoff. ACI is their opportunity to keep the high margins on the same equipment. Have you noticed how they are looking to keep the buying relationship with the network silo within the enterprise.

      Cisco has been a great partner to the network team in the past. One could argue so strong to a fault. Many decision makers just look to Cisco to tell them what new box they should buy within their stack. I’ve seen organizations that could get buy with stackable switches purchase 7000’s.

      If Cisco can bring the solution that meets the needs of their enterprise customers, they may still be able to dominate in the enterprise which doesn’t have the same pricing issues as the cloud players.

      1. I agree – the average enterprise doesnt have the same level of concern over pricing as infrastructure is not their core business. But at the same time, if private clouds run on gear that is substantially less efficient and substantially more expensive than cloud providers – that seems like a recipe to ask to have your IT department outsourced. Enterprise will never run at the same level of efficiency as a cloud provider, but it doesnt need to be night and day different, they can get pretty close, much closer than where they are today. While Cisco has been dominant, all of their efforts to get more into application oriented networking have failed from their AON efforts to their SONA efforts etc. Wasnt that long ago they gave up on their struggling L4-7 businesses and teamed with Citrix. As much as Cisco has made a lot of profit from talking about application centric networking, they have never had much success in that realm in practical terms. I was watching Facebooks head of networking talk at tech field yesterday who made a point to highlight how they werent simply trying to solve problems from the server or from the app or from the network, they were looking at the entire stack and finding the best places to solve technical challenges without an agenda – and it seems they are finding that many things some think should be done by the network shouldn’t, and perhaps the other way around too. We have talked about Cisco’s business challenges but I think they have a big one from a technology standpoint too – when I worked at Cisco they sent everyone a card with their employee badge with ideals and goals – on one side it said “no technology religion” and then on the other side it said “the network is the platform” – and to me I think that is a big problem for them, they have lots of bright young innovators that they encourage to use the network-centric-thinking to solve all sorts of problems pushing thinking into a box and leading to many technologists who believe they are doing the best thing not realizing they have been pushed down a certain path.

    2. There are a few things that were not mentioned in the article and comments:
      1. enterprise isn’t cloud: in the enterprise (virtual) machines are ‘pets’, in the cloud they are ‘cattle’. In the enterprise VM, OS and application are tweaked by hand. In the cloud things are automated. When a VM is lost, they automatically (autoscaling) start an other.

      2. sell into the other team: Companies where networking and server teams are silos. Vendors will sell networking products to the server team, because the networking team wouldn’t buy their products.

      3. cloud provider networks aren’t delivering on the SDN promise just yet. Because cloud providers connect servers with just a single connection. This isn’t a problem in a cloud provider network they have enough servers when a switch goes down and takes down 40 servers with lots of VMs they don’t care, there are thousands more which can fill the needs of their users. See point 1 as well. In the enterprise it’s different. They want all servers to have multiple connections to the network. Try combining that with SDN and things become a lot more complicated.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: