Brocade CEO says they've built an easy button for IP networks, are benefiting from SDN/NFV

easy button
Credit: Thinkstock

Brocade is best known for its storage network products, but CEO Lloyd Carney, who has led the company for almost three years now, says Brocade is making great strides on the IP network side of the house by reducing complexity. The timing is good, Carney says, because the arrival of Software Defined Networking/Network Function Virtualization has buyers looking at alternatives to their legacy suppliers. Network World Editor in Chief John Dix recently stopped by Carney’s office for an update.

Brocade CEO Lloyd Carney

Brocade CEO Lloyd Carney

You’ve been at the helm now for just about three years and the stock has more than doubled, what do you contribute your success to?

The first thing I did was put a good team in place because it comes down to the people. Jeff Lindholm used to run Asia-Pac sales for us at Wellfleet. I’ve known him for 25 years. My head of marketing, Christine Heckart, I’ve known for 12 years. She was CMO at Juniper when I was COO. My general counsel is from Micromuse where I was CEO. I’ve known her for maybe nine years. And these folks complement people that were already here, like Parviz Ghalambor, who runs Engineering for us, and Jason Nolet, who is head of Product Management on the IP side. So we have a good mix of old and new.

And we’ve shown our IP chops. We’re growing our data center switching business 24%-25%, while the market is only growing 8%-9%. If you asked someone three years ago about Brocade in the IP business, they wouldn’t even be sure we were in that market.

Financially, we’re doing well. We’ve raised the dividend. Our operating margins are around 25-26% and free cash flow is around $400 million. We’re on pretty good footing. We have a good portfolio of products. IP storage is a great opportunity for us. The Flash array guys are a great opportunity for us. The traditional SAN product is moving well.

How does the business break down today?

Around two-thirds is on the SAN side and a third is on the IP side. Besides the data center IP business we have the software and campus IP businesses, and if you combine all of that IP it is growing about 14%.

What’s growing the fastest in the IP portfolio?

The VDX, which is our IP fabric. One of the things we did -- and I can’t take credit for this because they did it before I got here – was take a SAN fabric and basically put an IP front-end on it, so it’s the same chipset but now it does IP. The thing people like about the SAN world is the ease of deployment and ease of use. You plug it in, it auto-senses what should be connected to what, auto load balances, and there you go.

That appeals to a lot of customers, especially if they’re making a bunch of configuration changes, 200-300 strokes per machine. Now it takes four strokes. We basically put an easy button on the IP network and that has been going really well. Instead of needing ten network engineers you need five. Instead of taking three days to stand up a machine you can stand it up in two hours. That ease of use has really gone over well.

Does the growth come from your installed base or are you winning new customers?

We’re winning new customers every day, but keep in mind our installed base is large because we have 70% of the SAN business. Ninety percent of the Fortune 2000 use Brocade products. It’s tough to find a customer that doesn’t have some of our kit, especially because we partner with so many companies. There’s no network company you can go to that has better partnerships than we do on the compute and storage fronts. On the storage side EMC, Hitachi, NetApp and Fujitsu all resell our product with their names on it. And on the compute side IBM, Dell, HP, Lenovo, you name them, even Hauwei, sell our products with their name on it. So it’s very difficult to find a customer who doesn’t have something of ours. They might not know it, but we’re in there.

Fabrics were really hot a few years ago, but then the hype died down, but it is still working for you?

Our competitors are making fabric-like architectures, adding easy button-like features to their product so it’s almost the norm now. You can either buy a fabric-like architecture that’s easy to deploy or you can buy one where you have a lot of open APIs you need to use to fully take advantage of the architecture. But a lot of customers are saying, “Just give me a flat Layer 2 connection because I’ve got 20 servers and 10 virtual machines per server, so I’ve got 200 VMs and I’m not sure what VM I need to go across what switch and it just takes too long to configure.”

In a fabric it doesn’t matter, you just plug them in, the network becomes transparent and so you can move those virtual machines around pretty seamlessly. So more people have evolved towards designing networks like that and some have stopped calling it fabric because it’s just what the network looks like.

Juniper made that big push with its QFabric. Were you there for any of that?

I was there at the very beginning of it. The challenge with that is complexity. When we launched the Brocade fabric we had 15 years of experience. Juniper didn’t have that SAN experience and tried to create this fabric from scratch. They took a complex ASIC from their routing module and that was the front end and you needed a dedicated, complex port on the server, not a regular IP port, to make the thing work. It was a headache.

Does your fabric appeal to a certain size/type of company in particular?

We have won across the spectrum -- universities, financial institutions, you name it. But generally mid to large companies. One of the top four automotive companies in the world. One of the top three banks in the world. We have another product in the works that sits on top of the VDX and has open API tools. It has all the programmability down through the architecture you could ever want. It’s built for customers like the high frequency trading guys that are doing really complex infrastructure to drive efficiency.

As a matter of fact, hosting providers are some of the best customers for the VDX because if you’re a hosting provider your network is your revenue engine. You’re selling ports to customers and with the VDX you can better utilize ports because they’re active/active. You run them as hot as you want, whereas in a traditional network it’s active/passive, one port is up and the other is a standby port that only comes on when the first one goes away.

And if you need less ports you need less space, you need less power, you need less cooling. So if you are a hosting provider and have your checklist of all the things you worry about, we’re probably going to influence eight of the top ten things in a positive way. You need less people also because it’s not as complex to configure these networks. Rackspace is a great customer of ours because they sell ports to customers. So they look at the model and they go – “Okay, this makes sense.”

Speaking of service providers, what is your customer mix these days?

In the SAN business, where the majority of Brocade products are sold through OEMs, we don’t have complete visibility into which sectors we’re being sold into. For IP networking products it is 45-47% enterprise, 40-42% service provider, 12-14% US Federal.

What’s your take on Software-Defined Networks and Network Function Virtualization (NFV)?

You can say there was a little irrational exuberance in the beginning, with everybody jumping on the SDN/NFV bandwagon, ourselves included. We bought a company called Vyatta before I arrived that is a premier product in that space. They’ve had millions of downloads and we’re well aligned with all the early adopters. You can go on Amazon Web Services, for example, and buy a virtual router for five hours if you’re doing some kind of test, and that’s our virtual router sitting there. Same thing with IBM’s SoftLayer.

AT&T is using our products because it scales really well because it was designed that way from scratch. Every other major NFV player took software from a physical device and squeezed it onto an x86 machine, so they have basic limitations because the software was originally designed for purpose-built ASICs. Vyatta built an architecture optimized for Intel. The first version of the product did 1GB packet forwarding and the next version with minor tweaks did 10GB of packet forwarding. Earlier this year we did a test with Telefonica Spain using a quad-core machine and did 80GB of packet forwarding using the same software because the product was designed to scale across multiple cores. Nobody else can do that. Nobody has the flexibility. We have, without a doubt, the best product in the SDN/NFV space of any major player because it was built for Intel, which is where the market is going.

We think the early adopters have gotten it and we’re engaged with them. Some customers jumped in and realized it is kind of hard. You need a certain caliber of engineer to do this.

We’ve made it easier but you still need specialized engineers. Some customers have that skill set and can do it. SDN/NFV for the average customer, it’s probably still early. Service providers are the early adopters. They have that expertise and they have the most pressing need. They have to change their cost model and the way to do that is get is rid of that dedicated, expensive hardware and go to x86 machines and cheaper software.

I know you just launched your own SDN controller.

Because our traditional switches and routers are OpenFlow enabled, it actually pulls the controller along. It’s a way for customers who want to try SDN/NFV to do so without a really complicated rollout. You can use the controller to manage the router and switch ports in your network. It’s a good entre point for customers because it is OpenStack-enabled, multivendor and it’s pretty easy to deploy, so it gives you a first step into this whole SDN/NFV space. Even AT&T is using it.

Some observers asked why we did a controller when there are a lot of controllers out there already, but customers said they want to buy it from one vendor and there are features they want that the open world will get to but some people need it sooner rather than later.

Do you view VMware with its NSX network product a competitor?

They’re actually a great partner. Two weeks ago I was in Germany meeting with one of my guys and he was late and I asked why. He said he went to meet a new customer about our IP fabric and I asked how he got the lead. He said the customer said, “The VMware guys were here and we asked what’s the best underlay for the NSX architecture and they said Brocade, so we’re calling you up.”

I’d be lying if I said every VMware shop needs us. For most VMware deployments the underlay doesn’t matter, whether it’s Cisco, Juniper, us, or anybody else. But if they’re going up against a Cisco ACI stack and Cisco wins, VMware loses. That’s a customer that’s never going to roll out NSX because they bought the Cisco version. In those instances they bring us in and our engineers work closely with them.

VMware is also a great customer of ours. They put up a new engineering building in Bangalore, that’s all our stuff. They’re redoing their data center here in Silicon Valley; that’s us. Their new Executive Briefing Center, it’s going to be our equipment.

If you’re a VMware organization and you have to create and tear down VMs all day, it would take weeks to go in old school and configure every port to say this VM needs to talk to that port over there, etc. With our easy button you plug it in, we acknowledge the virtual machine, the connection it has, and if it wants to move somewhere else to connect to some storage on here, we automatically do it. There’s no configuration required.

For a VMware engineer we’re like a God-send. This makes your life so much easier at scale. So we’re now getting to where VMware sales go-to-market guys are realizing that the underlay does matter some of the times. Where it does matter, we’re a much better partner than anybody else they have and we’re certainly a much better partner than Cisco, which is determined to squash them.

Speaking of Cisco, are you seeing much demand for their ACI stuff?

It’s an upgrade cycle with some of their customers. If you have Nexus 5000s or 6000s and want to run ACI you need to upgrade the switches. I hear a lot of customers complaining about that. I’m not hearing about a lot of big wins.

Has the arrival of SDN/NFV changed the field of competition?

I think it’s caused a lot of customers to rethink their architectures, because they know they want to get to SDN/NFV. Quite frankly, if I didn’t have an SDN story I would not be growing my IP revenue in the mid-teens because customers might not be ready to go there yet, but they want to know you can take them when it’s time. There’s a 10-step path. Some customers are at step one and then you’ve got AT&T at step 10, deploying, running hard. A lot of customers are in the middle and they know that if they buy a switch from Brocade today it’s SDN-enabled. They’ll be able to use SDN to manipulate that device. That has been great for us. It’s probably the single biggest reason why a customer who didn’t think about Brocade before thinks Brocade now.

Let’s shift over to the SAN side of the business. What’s new there?

1 2 Page 1
Must read: Hidden Cause of Slow Internet and how to fix it
Notice to our Readers
We're now using social media to take your comments and feedback. Learn more about this here.