olly - Fotolia

Evaluate Weigh the pros and cons of technologies, products and projects you are considering.

VMware's NSX use cases are missing in action

Ask network admins why they think NSX use cases are few and far between and this is what you'll find.

Normally, I don't blog on flights home from trade shows. I'm not elite enough to reliably upgrade, my legs hurt, and I'm not willing to play reclining seat versus laptop chicken. Dequeuing email on my phone is usually productive enough. But today on the way home from VMworld, I've been staring at the seatback ahead of me, mulling an uncomfortable discovery: In dozens of conversations at the show, I only found one team using NSX in production, while others are paying for shelfware. And administrators hate shelfware.

Biting the hand that feeds me

Full disclosure: I don't really believe anything until an admin says he or she is doing it in production. It's easy to talk the talk, but bullhockey walks when it's time to support business services on the wire. And on the first day of the show, a customer asked about a previous article I wrote on data center NSX use cases. He said, "I know you're a fan of SDN, but what's a typical use case for NSX?" Before I could reflexively rattle off a few, I reviewed them in my mind against Step One of crawl-walk-run engineering best practices and, subsequently, threw the majority away as unlikely big-bang projects. For the remainder of the show, I found myself asking any hardcore VMware admins I came across, "What's your networking use case for NSX?"

But NSX seems stalled -- especially in midsized enterprises -- and hopefully a bit of feedback helps move it forward.

Before I net out the results of that informal poll, let me first say that I'm a huge fan of VMware. I was a very early adopter of both Workstation and then ESX and could not do my job without vCenter. I've managed Hyper-V, Citrix, CoreOS and cloud, and I still enjoy VMware. I'm also a fan of programmatically managing All The Things, and I won't be sad to see CLI relegated to the back of the drawer -- something we never really part with, but don't think about very much. But NSX seems stalled -- especially in midsized enterprises -- and hopefully a bit of feedback helps move it forward.

Top five list for not having NSX use cases, late night virtualization edition

With that, here are the most common reasons given by technology influencers at the show for not having any NSX use cases:

  1. "It's too hard to deploy, even in the lab." This was the most common administrator complaint. With a guru and a book of spells, I got it running in our lab, but not without a cut finger and heavy EMF exposure. For a virtual product, there sure seems to be a lot of hardware and cable involved. True, once it's running, it's easy to manage, but you can say the same thing about Cisco 1000Vs. Without a quick-config process and how-to support, too many admins are walking away from proofs of concept because they just don't have time.
  2. "My boss wants it, so therefore I'm skeptical." Admins don't like having products pushed on them. We like to pick them ourselves. When vSwitch monitoring and NetFlow export stops in the wee hours, we seek trustworthy over shiny. Practitioner-focused education would nurture users and boost demand.
  3. "It's way more expensive than Cisco/Juniper/Brand X." At thousands of dollars per CPU, or hundreds per year, per VM, NSX can be a budget stick-out. However, beneficial, jargon-laden feature descriptions can make it tough to get management on board or, in some cases, even renew initially discounted licenses.
  4. "My management is too risk-averse." As admins, we've concentrated so many services on so few boxes in the last 10 years. Management is happy about the cost savings, but increasingly concerned about systemic failure of the fabric. Replacing core data center infrastructure with anything out of the ordinary is a tough slough.
  5. "I have more important things to do." In other words, IT folks don't see enough value to invest in it. For many engineers, the physical hardware they already have, especially with newer, high-performance ASICs, is working just fine, thank you very much. This is an evangelism issue for the most part. Benefits are numerous -- for example, distributed firewalls, which are especially huge for east-west traffic -- but VMware needs to get a bullhorn and get the word out to the teams that push packets.

Still time for answers, but competition is mounting

Of course, there are some other macro challenges impinging NSX use cases -- factors like an anti-SDN-bias, less developer support than Cisco has for ACI with DevNet and the prospect of open software options on the horizon -- but these aren't anything VMware can't address. With enormous resources at its disposal, VMware can easily boost its investment in education and evangelism. Revisiting pricing wouldn't hurt either.

That said, the "I have more important things to do" argument will only be addressed when it actually becomes a priority to hand over our network keys to our virtualization platform vendor.

There's still time for VMware to raise awareness and get everyday admins excited and snapped into vCenter for network management. But not much -- 21-inch racks are beginning to circle.

Next Steps

Answering some NSX use case questions

What's next for NSX?

Beefing up VMware, DevOps stokes use cases

This was last published in September 2015

Dig Deeper on Data Center Networking