How Brocade totally missed the boat with Vyatta

When Brocade acquired Vyatta, it didn’t seem that they fully comprehended just what they had their hands on. Vyatta was an efficient, powerful, flexible network operating system based on Linux, which could run just as easily on commodity hardware or (para)virtualized infrastructure. The possibilities were endless. Unfortunately, the result was predictable.

The concept behind Vyatta was amazing

Think about it: why spend the price of a mid-range supercar on a top-tier brand (with only the processing horsepower of a modest cell phone) when you could handle 10+ gigabit of filtered traffic and millions of packets per second for a few grand?

Maybe you need to quickly provision a full-featured router for a lab before that big demo on Friday. Don’t pick up the phone and call your sales rep. Instead, just drop a Vyatta image on your existing Xen/VMWare/KVM/whatever infrastructure and plumb the interfaces. Could be that you’re tired of fighting with Amazon’s integrated VPN facility and want something that just works, and works the same everywhere you need it. Vyatta can be that, too. Want a real home router that’s not designed to be used by somebody as technically inept as your grandmother and doesn’t require daily reboots? Yeah, Vyatta again.

“But what if you need to add a captive portal for school library visitors with a filtered proxy and a district-mandated anti-virus gateway for normal Internet access, except on every 2nd Tuesday of the month, where all image search results must be replaced with motivational posters from the 1970’s featuring cats in humorous poses?” I hear you ask. Well, first, that’s a hell of a run-on sentence. You should be ashamed.  Secondly, Vyatta’s based on Debian, with a full user-land, so you can run pretty much anything you want on there. Your imagination’s the limit.

Open source, not secret sauce

The real beauty of the whole concept is that Vyatta is based on a bunch of open source tools, quilted together nicely in a neat stack and administrated from a unified, Junos-like interface. This means all Brocade had to do was lead the project, nurture their community (or at least try not to piss them off too badly), occasionally push bugfixes and new features back upsteam, and reap the rewards. Much of the heavy lifting is done by developers who aren’t even on their payroll.

And with open source comes the associated community; essentially a loyal, dedicated user base. While some companies view these folks as leeches who just want something for nothing, even from a purely commercial standpoint the reality is actually a lot more interesting. You can generally categorize a commercially-driven open source community into three main camps:

  1. Developers contributing code either because other interested companies are paying them to, or because they like the product and want to improve it for their own use. In other words, free code.
  2. Users who enjoy the product, understand its benefits, and eventually go on to become paying customers. These are also quite often the same folks offering ad-hoc technical support to their peers, thus alleviating some of the load on customer service. Again, for free.
  3. Those who will never pay for the product still add value indirectly by acting as evangelists. To put it another way, free marketing. (And if they’re using your product without paying, at least they’re not cutting any checks to your competitors, either.)

 Missed opportunities (could’ve, would’ve, should’ve)

Brocade really could have done a lot with the platform. When I heard they’d acquired Vyatta, I’d personally envisioned Brocade-branded core switch-fabric with dozens or hundreds of ports running Vyatta under the hood. Powerful routers with all the flexibility and control you’d expect from your Linux servers.

Or, imagine SOHO network gear that’re designed as open platforms: inexpensive, running open source software, mod-friendly, easy to cobble into something bigger, more interesting. Budget-conscious and enthusiast users tend to love platforms which are flexible, extensible and aren’t limited by the feature list that tickles the product lead’s fancy that particular quarter. And like it or not, those people make up a huge chunk of the overall market.

As a case in point, consider the legendary WRT54G series home routers from Linksys. They sold countless millions of those little guys, with a run that went on for 10 years solid. It wasn’t the hardware that was so special, but the fact that the platform was extremely hacker-friendly. A product like that would have let Brocade grow into a whole new market space, and Vyatta could have been the linchpin.

If you could run the same network OS you’re already familiar with in your public cloud infrastructure, your colo, your office network, your lab, your home… exactly which market wouldn’t they be dominating a few years from now? The price to performance ratio of solutions like those should have left the likes of Cisco and Juniper wondering why their analysts never saw it coming.

The predictable outcome

Rather than innovating, Brocade has consigned Vyatta to a proprietary VM appliance for use only in the mythical cloud, tucked away in the backwaters of their corporate website. The Vyatta forums are closed, open source builds discontinued, the community discarded.

RIP Vyatta

Perhaps Brocade was afraid of undercutting their [arguably stagnant] switch market. Maybe they simply didn’t know how to sell it, exhibiting the same lack of imagination as Vyatta before them. In any case, the sad truth is that in my opinion, the unique opportunity presented by the acquisition seems to have slipped through their fingers.

Vyatta is dead, long live VyOS / EdgeOS

Perhaps the best outcome of this unfortunate eventuality was the resulting fork. It seems Ubiquiti picked up several Vyatta expats, creating their own commercial/open-source fork dubbed EdgeOS. They currently have several platforms in the EdgeMax line which can filter millions of packets per second for well under a grand. I haven’t yet got my hands on the hardware to test it, but knowing the underlying software platform, this is an extremely interesting development for those looking at routers in the SOHO and SME market spaces.

As for the open source community, they’ve created a fork of Vyatta called VyOS, picking up right where Vyatta left off. So far, the momentum is encouraging. Many of my own bugs have already been acknowledged and either addressed or are scheduled to be fixed in the next upcoming release. VyOS is also presently backwards compatible with the last community release of Vyatta (6.6R1), so current users of Vyatta are not left without an upgrade path.

For my part I’m still hoping that, against all odds, Brocade will realize how badly they’ve fumbled and will eventually recognize the potential they’ve missed up until now. A world where powerful routing hardware runs on a flexible, open source software platform is one I’m very much interested in experiencing. Until then, I’ll be doing my best to support and contribute to both EdgeOS and especially VyOS as much as I’m able. Even if it’s too late for Vyatta, here’s hoping for a long and fruitful life for the legacy Vyatta has left for us.

13 thoughts on “How Brocade totally missed the boat with Vyatta

  1. Pingback: Leaving pfSense for VyOS (formerly Vyatta) | DotBalm.org

  2. Thanks for the information, we use to be a cisco shop, but started using vyos a few weeks ago and we are very impressed and like vyos a lot. Keep up the great work. We will continue to support vyos. Great firewall / router appliance.

    • Good call. I just implemented a new pair of EdgeRouter Pro’s to replace an underperforming Cisco router for my company’s corporate network a few weeks ago. I think the improvement was more drastic than anyone expected.

  3. Sad to see this. Anyway to know how many Vyatta routers/FWs still being used roughly? or how many people are using Vyatta?

    Thanks,

    –Frank

    • No idea, though I imagine Brocade has some idea if people are still trying to hit their mirrors. I moved my own routers to VyOS some time back without difficulty, and I’ve recently setup some EdgeRouter Pros in an office environment. From an end user perspective, one could argue that the situation actually got better, since we have more options now.

  4. As a former Vyatta employee I can tell you that Brocade never “got” opensource. Frankly, they never understood what they bought. Brocade bought into the market positioning that Vyatta was SDN company – and they wanted some of that magic pixie dust to rub off on them, But Vyatta, though software, had no integration into SDN controllers, no separation of data and management planes . . . . it is a router/firewall/vpn. That’s it.

    Instead of trying to rally the enthusiasm and skill of Vyatta users and selling service and support, Brocade killed the opensource disti and made it a not very good Juniper wantabe.

    • I’m glad you mentioned SDN. I started writing about SDN in the original article, but quickly found myself in a tangent. In other words, I could have easily written as much about what Vyatta wasn’t as what Vyatta actually was. To date, it seems like at least Ubiquiti does fundamentally understand the latter, as evidenced by EdgeRouter and their new EdgeSwitch line. I hope it’s some consolation that all your hard work wasn’t in vain.

  5. We replaced a Cisco 1800 series router that kept shutting down ports when we got busy with Vyatta on an old PC and have never looked back. We built a rack-mount dedicated system for it recently, a micro-atx atom based MB with two gig nics built in, a 32G SSD for the image and a nice little case from newegg and it is doing a better job than any of the commercial routers that people have been trying to sell me. We upgraded it to the current VYOS release and it performs as well as anything I could purchase.

    • Sounds like a wise approach. The only thing I’d add to that is make sure to get regular backups of your configs; low-end commodity gear can be a little flaky.

      Also, feel free to use the VyOS mirror I’ve recently built: http://mirror.tuxhelp.org/vyos/
      It’s on a fast uplink and optionally supports IPv6 and TLS. I find it preferable to the default mirror pool, since that could potentially link to any VyOS mirror anywhere in the world, any of which might or might not be available at a given time.

  6. I’ve used Vyatta in all the companies I’ve worked for and also at home. At one place we build the whole network on Vyatta, 10Gig Intel NICs and commodity hardware and years later it is still going strong.
    In some cases move to VyOS wasn’t successful and from what I hear it’s mainly because of some Quagga issues and the fact the most of my routers had lots of BGP peers, but I guess this is about to change.
    Anyway I’m really sorry that the lack of common sense in Brocade killed the project but am confident that the community will remain strong and VyOS will develop further.

  7. The VP of Platform Engineering at http://www.wikia.com tells me their origin has been served entirely by Vyatta/VyOS since early 2009. They’re currently serving ~3 billion pageviews/mo. and 35k http req/s for over 140 million users world-wide. What brocade did here was disappointing. The good news is how the community has flourished with the VyOS fork and Ubiquiti. Even though EdgeOS is still behind the current stable VyOS it’s a ridiculously capable/affordable option for corporate campus routing for everything but the largest companies. It’s also used in all of Wikia’s offices around the world.

  8. Just came across this post now and do agree it is sad, VyOS is now my preferred edge gateway for disparate networks. Just finished deploying a site to site VPN using VyOS on virtualised server tin.

    I have been keeping a keen eye on the opensource CloudRouter project and looks promising, however the documentation is still getting there and not many guides and builds made public, but am hoping it to soon be able to replace many Vyatta/VyOS variants.

  9. Well, three years later and this article has been properly vindicated. Recent events have more or less proved two things:

    1. White-box networking is officially a thing now. Naysayers who claim you can’t route a lot of traffic without Cisco silicon should talk to AT&T about it, as they plan to run 55% of their network on such by the end of the year, climbing to 75% by 2020.
    2. Apparently Brocade never really did get Vyatta, and never truly did anything with it. Probably at least partially because it wasn’t really the SDN they thought it was when they made the purchase, but also because they had a certain mindset about undercutting their existing product lines.

Leave a Reply

Your email address will not be published. Required fields are marked *