Home > Cloud Computing, Virtualization > Can The Success of Server Virtualization Be Repeated in Networking?

Can The Success of Server Virtualization Be Repeated in Networking?

The software-defined networking is the new buzzword for network centralization, which is also known as OpenFlow or network virtualization. The idea is to centralize the control to a server (or a cluster of servers) called controller.

With the acquisition of Nicira by VMware, the software-defined networking has caught many eyeballs from the community. From there, VMware extended it to a new vision called software-defined datacenter which includes three elements of computing: compute, network, and storage.

Lost VMs or Containers? Too Many Consoles? Too Slow GUI? Time to learn how to "Google" and manage your VMware and clouds in a fast and secure HTML5 App.

With VMware’s dominance in server virtualization, the software-defined datacenter presents a new opportunity to grow into new areas like network and storage. Because of the acquisition, most people believed VMware had moved Cisco’s cheese and there would be inevitably some conflicts between the two companies moving forward.

Which is more important, Software or Networking?

Technically speaking, the software-defined networking has two aspects: one is the defining software, and the other is the networking. VMware and Cisco both have its winning strengths. From high level, Cisco for sure knows better about networking which it has helped to grow from its very beginning; VMware certainly knows the software better as demonstrated by its success in server virtualization.

The question is – can they repeat their success from either networking or virtualization into the new area? Of the two aspects, which one is more important as a success factor?

What’s Different This Time in Networking?

The success of server virtualization is mostly to the credit of VMware which had picked up the technology from IBM mainframes and developed it on the dominant X86 server market. Looking back at history, I think it’s also partially to the credit of Microsoft. It’s not because of the competition from Hyper-V that moves the market faster, but because of the hardware abstraction and ecosystem Microsoft had successfully created. It may not sound straight-forward, but let me explain it a bit more.

In the early history of computer, the hardware and software (and more often than not, services) are all bundled together. IBM chose Microsoft to provide DOS for its PC. Microsoft grabbed this opportunity and created a new layer above hardware but below application, and successfully marketed it as a separate product in the whole stack. After this, PC industry has moved to clear layers with high specialization: hardware by IHVs like HP, Lenovo, Dell; Operating Systems by Microsoft and later Linux; middleware and application by ISVs like SAP.

With this layered structure, Microsoft made huge money. It also opens a door for the later virtualization – in some way hypervisor layer is very much like OS. Having one layer (OS) there helps another layer (hypervisor) to sneak in. Without the OS layer there in the first place, it would be pretty hard for a smooth adoption of VMware virtualization. The business model for VMware to market its vSphere is very much the same as Microsoft markets its Windows – they talk to the same IHVs for huge OEM deals.

In networking world, we don’t see a clear separation of layers that are owned by different vendors. Most networking companies sell boxes bundled with hardware and software. There is no “Microsoft” in play as it’s in compute stack. Inserting an abstraction layer by another vendor would be extremely hard, if ever possible.

What is Going to Happen?

As I mentioned in the previous post, the niche that SDN can play is with large scale green-field datacenters. There will hardly any player can dominate the high value “controller” market in the same way as Microsoft does with OS and VMware with hypervisor. The concept of network stack is simply not there due to various reasons.

The technology will be quickly adopted by networking vendors who will also develop the controllers by themselves or acquire startup companies. Once the technology is mature enough, these networking vendors will bundle them with existing products and gradually market them to existing customers.

For virtualization software vendors like VMware, they still have places in the game. Because they mostly own the virtual networking, they can apply these SDN related technologies over there. It remains to be seen whether they can break the virtual boundary into physical networking.

  1. May 6th, 2013 at 08:31 | #1

    Can The Success of Server Virtualization Be Repeated in Networking? http://t.co/Ia0NRyXL8H via @sjin2008

  2. May 6th, 2013 at 09:12 | #2

    Can The Success of Server Virtualization Be Repeated in Networking? (DoubleCloud) http://t.co/uHPRG9a0Tk

  3. May 7th, 2013 at 10:11 | #3

    Can The Success of Server Virtualization Be Repeated in Networking? – http://t.co/ES8bL8j8n8

  4. May 7th, 2013 at 19:45 | #4

    .@sjin2008 asks: Can The Success of Server #Virtualization Be Repeated in Networking? http://t.co/0jZumJBtXU #SDN

  5. May 23rd, 2013 at 05:45 | #5

    Really a good post with interesting considerations.
    Also networking design is changing moving from L2 to L3 (for example with VXLAN or in spine/leaf architectures) and more control and abstraction could be needed.

  6. May 26th, 2013 at 13:52 | #6

    Thanks Andrew,
    That is true. I think I need to learn a bit more VXLAN. :)
    Steve

  1. May 6th, 2013 at 16:59 | #1