Jayshree Ullal, Arista Networks | VMworld 2015

Jayshree Ullal, Arista Networks | VMworld 2015


>>Cisco, extracting the
signal from the noise. It’s theCUBE, covering vmworld 2015, brought to you by VMware
and its ecosystems sponsors. Now your hosts, John
Furrier and Dave Vallante.>>Okay, welcome back everyone. We are live here in San
Francisco at Moscone North Lobby. This is SiliconANGLE’s
theCUBE, our flagship program. We go out to the events and extract the signal from the noise. I’m John Furrier with my
co-host Dave Vallante. And our next guest is Jayshree Ullal, The President of, CEO of Arista Networks. Welcome back to theCUBE. We haven’t seen you in
couple years, welcome back. You look great.>>Good to be here, John, Dave. I see you don’t put me
in the middle anymore. (laughter)>>I know, we want to stare right at you and get all the data
from out of your head, and get it, share it with the audience. Well, first thing I want to
say is last time we spoke, you were a private company,
now you’ve gone public, IPO. Congratulations. What’s it like? What’s it like from private
company to public company? Share the experience.>>It’s definitely different, for starters we’re not Arista Networks, we’re ANET. We are a four-letter symbol, I guess. So abbreviate everything. And then people just
track us a whole lot more. And, you know, there’s
an automatic branding, an awareness of the
company, and anything we do, every time we sneeze,
we get written about. Good, bad, or not.>>You guys are pacing the
market, and I remember, Dave and I, when we first started theCUBE, we were in the Cloudera office, and then when we first chatted, we’d see the boxes of Arista coming in. You guys made a great mark early on around people doing large
scale, lot of networking. But the market’s changed. SDN has exploded, VMware bought
Nicira, SDN’s the hot thing. NSX is doing well, as Pat Gelsinger said. What’s going on? You guys have done some things. SDN certainly is, takes the
market to where you guys had originally had your vision. What’s the update with that whole SDN and how does Arista play into that?>>I think if you step back and
look at SDN in the beginning, there was a lot of confusion. And my favorite acronym for
SDN is Still Don’t Know. But I actually think we still do know now, and we’ve gone from it
being a marketing hype to really about openness, programmability, and building an infrastructure to do network management correctly. Software clearly drives our industry, and more importantly
drives capxn OPX reduction. And what’s really happening is there’s a lot of change
where it’s not just devices and users and traditional applications, but really it’s about
workloads and workflows. And if you can realize
there’s so many different types of workloads that need control, and so many different types of workflows that need telemetry, that
fundamentally is the essence of SDN in my view, and
it takes a whole village. Arista can’t do it alone. We’re doing a lot of things
on programatizing our stack and making network more
open and programmable, but we work with a whole slew of vendors to really make it possible.>>During the early days,
open flow was the buzzword, came out of a lot of
academic stuff that was being what the geeks were working on. What do people get right? And there was lot of missteps
early on with open flow, and only because it’s early on. What did SDN get right,
or did they get it wrong? And how did you guys see that ’cause you guys were
already out shipping product when this hit. So what’s your observation
of what went right, what didn’t go right,
what’s going right now, can you share your insight?>>Yeah, I think, you
know, our founder Ken Duda would say this very well, which is when you look at open flow, it’s a little bit of a technology
searching for a problem. When you look at what Arista did with our extensible operating system, we built a state-oriented,
publish subscribe model to solve a problem. And the fundamental problem
we were solving was, we saw the industry building
monolithic enterprise stacks when everybody was moving to the cloud. What are the three
attributes a cloud meets? You got to be always on, you
got to have scale, right, and you have to have tremendous agility. You got to move across
your workloads fast. And that, to me, is the trick behind SDN not latching onto a technology, but whether it’s open
stack or big data analytics or new cloud applications or bringing the LAN and the WAN together, or places in the network converging, fundamentally, we were
cloudifying everything whether it’s public, private, or hybrid.>>So I got to ask you, I know
you’re going to see Pat Gelsinger shortly after this interview. Two themes that are coming on
the queue over the past year around networking has been
resiliency and agile, agility. Those two factors ’cause you have vertical and now horizontally
scalable things going on. What’s your take on that? As someone who’s been in
the industry, you’ve seen kind of the old generation now transition to the new generation,
cloudification, API-ification, these are are new dynamics that are table stakes now in cloud.>>No, they are. And yet, if you look at the, both problems are hard problems. They cannot be solved by
sprinkling some pixie dust. And what I mean by that is when you look at something
like high availability, in the past in networking,
you had two of everything, two supervisors, two operating systems. You had something called
in server software upgrade, so that you’d bring one down
and then bring the other. But today, there’s no tolerance
for two of everything. You know, no customer wants
to pay for two of everything even if the vendors want it, right. So what you really need
is smart system upgrade where you’re doing everything real time. You know, at the colonel level, you need to automatically
repair your faults. Software has memory leaks. Software has faults. It’s how quickly you diagnose
them, troubleshoot them, trap them and recover from them. And then if you look at hitless upgrades, you got to do them real time, you can’t wait to have
an enterprise window and bring it down and bring it up. Your boot time, your convergence time has to move from minutes to seconds, and the biggest thing you have to do is, let’s look at simple
command like copy paste. We do this over and over and over again. Change control has to improve. Rather than doing it every
time, a hundred times, wouldn’t it be nice if
you could just press one command and it happens
across the entire switch, across all the ports,
across the entire network. So I think the definition
of high availability has completely changed
where it’s really about network rollback, time
stamping, real time recovery, and not just two or three of everything.>>So, it’s, tight time here with you. John mentioned a public company, you guys have beat five quarters in a row, of course, you know, you get on that slope and the pressures go. But you can’t fight the
whims of the market. You just have to execute, and you guys are executing very well. Great growth, you’re
clearly gaining share. Partnerships. You announced a deal with HP
in converged infrastructure. Just saw this week, or
maybe it was late last week, that HP is OEMing NSX. So now it’s got a really
interesting converged play with Arista against Cisco. I want to talk about the
competition and that partnership.>>Well, it’s not so much against Cisco. It’s following the trends. And I think there are
two major trends, right? And they’re actually C letters, too. Cloud and converged. So if you look at what
Arista’s really doing, we’re serving a big public crowd trend. We’re in six out of the
seven major cloud operators. And there’s no doubt that
the cloud is happening, it’s not just a buzzword.>>You call ’em cloud titans.>>They’re called the cloud titans. You’ve done your homework. Good job. And hopefully, I’ll be able
to come back to the theCUBE and say we’re in seven out of seven, but today we’re in six out of seven.>>And the cloud titan is
the big hyperscale guys, is that right?>>Absolutely, and we’re just in a very early inning with them. Everybody thinks we’re already saturated. We’re just beginning. How many innings are
there in a baseball game?>>Nine.>>Nine, in cricket there are only two.>>What inning are we in?>>No, we’re in the first. Of two in cricket, a long way to go. (laughter)>>Cloud Native’s right around the corner. What do you think of Cloud Native? What does Cloud Native mean to you?>>So, the Cloud Native really means bringing the cloud experience to public, private, or the hybrid. So you talked about the HP partnership. And over there, it’s not
really building a public cloud. It’s about bringing a private cloud where you bring in the
compute, the storage, the virtualization, and the network as a converged experience. Now, that one we can’t do alone. And I couldn’t think
of two strong partners, better partners or stronger partners than VMware and HP to help do that for us.>>Well, you said it’s not against Cisco, but that’s a great alternative
for the leading products in the number one marketshare.>>Absolutely, I think
the enterprise companies have to have a wake up call. They need to understand
that the one neck to choke or one lock in that’s all proprietary is a thing of the past. And really, it’s about building best of breed building blocks.>>So I want to ask you,
just on some current events, and I’ll see buzzwords that
get recycled in every trend, is QOS policy-based fill in the blank. Everything’s policy-based now, so that makes a lot of sense, I get that. Apple just announced a deal with Cisco where they are throttling,
I shouldn’t say throttling, or deep packet inspection, I
won’t say those two things. (laughter) Giving iOS users a preferred
fast lane with Cisco gear, so it brings up this notion
that workloads are driving infrastructure or devops, if you will. What’s your take on all that? Are we going to see more things like that? Are we going to see more
customization around prioritization?>>Well, I think QOS and especially policy are definitely overused words. First step, I don’t
think you’d apply policy to an application to
make your network better. What you really have to
do is make your workloads and workflows go better and
have some control for them. So I’m not a big fan of
tweaking every application of the policy ’cause the
applications are changing, right? But if you look at what Apple’s doing, I think this is a great thing for Apple because what they’re really
doing is consumerizing and enterprising their
systems and devices, right. You’re seeing the convergence of consumer and enterprise coming together. So I see this is really about improving all of our iPhone experiences
across the enterprise.>>We got to wrap up ’cause you
got to go see Pat Gelsinger. But I want to ask you one final question. You’re an inspiration to the industry. You’ve been around a
long time, you know a lot and you’re leading a public company. What are the opportunities that
you see for folks out there, boys and girls, men and women,
in science and technology and in entrepreneurial opportunities?>>Yeah, I’m glad you ask this question because I think it’s too easy
with everything being hot for everybody to want to go straight to the top rung of the ladder. And I was telling Dave and you before, one step at a time. First you have to build your
foundation on education. Boys and girls, education is important. Follow your heart, follow your
dreams with math and science. You know, my dad started the IITs and he pushed me in engineering,
and I didn’t like it then but I realized you can be a cool engineer, and before Moscone got started, I actually went into the
manhole of every PG&E circuit to make sure that the
electrical circuits were okay for this now fantastic convention center.>>Can you help with the wifi?>>Back in those days, there was no wifi. That’s the next step. So I definitely say,
build your foundation, follow your dreams, but
go one step at a time. Don’t expect to be at
the top rung right away.>>I know you’re a parent. We are friends on Facebook. What’s your advice to
the younger generation in terms of opportunities
that they could pursue in science and math? There’s a lot more
opportunities, interdisciplinary, not just computer science
or electrical engineering, like it used to be when
we were growing up, but now it’s much broader. What are some of the things
that you get excited about?>>I get excited about science. I think when you look at engineering, it’s about applying science. You know, know your
fundamental math, science, you know, physics, chemistry,
bio, whatever turns you on. And don’t make an
assumption that it’s tough or hard til you’ve been through it. You know, I had seven years
of physics in high school. I don’t recommend seven for everybody, but, you know, but I didn’t
really care for biology. So I would say never shy away from trying something til you know. And then, of course,
there’s applied science, whether it’s computers or programming or media arts or visualization that you can add on top of that. So you’re very right. I think there’s the cake,
which is your foundation, and then there’s the icing where
you can build on top of it.>>And will they find their passion?>>Absolutely, find your
aptitude and passion. You know, you don’t try to
do drawing or needlework if you’re not good at it. I wasn’t. And I know my mom despaired about that, but you go, follow both
what you’re good at and what you’re passionate about.>>Jayshree, thanks so
much for spending time. I know you’re super busy. Congratulations on your successes.>>Thanks for having me here, it’s always a lot of fun.>>And we got to get you back on. This is theCUBE, bringing
you more signal here all the data here in the theCUBE. We’ll be right back, more
live from San Francisco after this short break.

2 Comments

Leave a Reply

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