Exaforge

Cloud, DevOps, Evangelism

Is NSX Really New?

I had an interesting request from a colleague recently.  They had a customer suggesting that NSX was really nothing new, and could be replicated more cheaply.  In the customer's words:

"NSX is a suite of tools (not marketed that way, but it is). OpenVSwitch is a single tool. With IPTables, StrongSwan, OpenVPN and OpenVSwitch as a collective suite of tools you can get complete NSX functionality for $0 on Linux."

As you might expect, I have a few thoughts on this.  The value of a suite of products is greater than the sum of its parts.  Is it possible to get similar functionality for a narrow set of requirements using iptables, OpenVPN, OpenvSwitch, etc?  Sure.  Is the cost of that $0?  Absolutely not.    Consider the following that NSX offers (as a suite):</p>

</span>

  • Full support from a vendor (non trivial, includes QA testing, and multiple experts to call when things go wrong).  Not something a home grown solution can offer.
  •  Full support for a huge range of operational models.  Not something a Linux-only solution can offer (as much as I love Linux and detest Windows, Windows does exist).
  •  Full REST API to integrate into a larger workflow (which is half the point of network virtualization, no?) - could be built by a lone guy, but now he's responsible for a huge dev project.
  •  Community and Vendor Ecosystem.  Want your stuff to terminate into the real world?  Arista, Broadcom, Brocade, etc can all help with that with NSX.  Homegrown?  Maybe - maybe not.

Time costs money - its not free.  By the time you built out the required software, made it resilient, added a good GUI, built integration into the common products out there, added a API and built a partner ecosystem, you would have built a company worth about $1B.  How do I know?  Because someone did - Martin Casado, and that company was Nicira - now owned by VMware and the basis for VMware NSX.