>> From New York City, it's theCUBE, covering Welcome to the New Edge. Brought to you by Pensando Systems. >> Hey, welcome back everybody. Jeff Frick here with theCUBE. We are in Manhattan at
the top of Goldman Sachs. It is a great view if you
ever get an opportunity to come up here, I think
43 floors over the Hudson. You can see forever. But this is a cloud event
so the clouds are here in. We're excited to be here. It's the Pensando launch. The name of the event is
Welcome to the New Edge, which is a pretty interesting play. We hear a lot about edge,
but we haven't really heard of a company really
focusing on the edge as their primary go-to-market activity in really thinking about the edge first. So we're excited to have the co-founder, Cube alumn and many-time
guest Soni Jiandani. She's the co-founder and
chief business officer. Soni, great to see you. >> Good to see you, too. >> And our host here at
Goldman Sachs is Josh Matheus, he's the managing director
of technology at Goldman. Josh, great to see you. >> Me too, thank you. >> And thanks for hosting us. Nice place to come to work every day. >> You're right. (laughs) >> So, great conversation today. Congratulations on the
launch of the company. Over two years in stealth mode. Talk a little bit about that. What it's like to be in
stealth mode for so long? And you guys raised big
money, you've got a big team, you're doing heavy-duty technology. What's it been like to
finally open up the curtains and tell everybody what
you've been working on? >> It's clearly very
interesting and exciting. Normally it's taken me nine
months to deliver a baby. This time it's been two and a half years of being in stealth while
we've been getting ready for this baby to come out. So it's phenomenally
exciting that, to be sharing the stage with our
customers and our investors and our strategic partners. >> Yeah, I thought it
was pretty interesting that you were launching with customers and when you really
told the story on stage, of how early you engaged
with Josh and his team. First I want to get kind
of your perspective. Why were you doing that so early and what did that ultimately do with some of the design
decisions that you guys made? And then we'll come back to Josh, as to, you know, his participation. >> So, I think whenever you
conduct technology transitions, having a sense from customers
that have the ability to look out two to three
years is very important because when you're
capturing market transitions, doing it with customer
inputs is far more relevant than going about it alone. The other key thing about
this architectural shift is that it allows the
flexibility for every customer to go take pieces of how
they want to bring the cloud architectures and bring
it into their environment. So understanding that use case. And understanding the compelling
reasons of what problems, both technological and
business, can we be solving. And having that perspective,
into the product definition and the design and the influence that customers like Josh have had, is why we are sitting here
and talking about them in production as opposed to,
yeah, we're thinking about, we're looking at it from a
proof-of-concept perspective. >> Right. And Josh, your perspective,
you said earlier today that, you know, as long
as Soni's involved, you're happy to jump in and
see what she's been working on. So how, you know, how
did you get involved? How did they reach out to
you, and what is it like working on, you know, technology
so early in its development that you could actually
have some serious influence? >> Well, it's an amazing opportunity to get exactly what you want,
exactly what you know is going to solve problems for the business here. And, you know, the other thing
is, you know, we've worked with this team through
almost every spin in. I think I was a little young
for the maybe the first one, but otherwise, this team
has worked with them through at least 15 years or more. So we knew the track record for execution and then, for us, on this product, I mean, it was an opportunity
because it's truly a startup. You know, Soni and the team brought us in, we kind of just put out
problems on the table that we were trying to solve. And then, you know, they
came up with the product and the idea and we were
able to put together, yeah, these are our priority one, two,
three that we want to go for, and, you know, we've just been
developing alongside them, both software and, you know, driving what the feature set is. >> Right. So what were some of those problems, at a guess, it probably
seemed like forever ago when you started this conversation, but as you kind of looked forward a couple of years back that you could see that were coming that you needed to address? >> You know, it's funny, we
started with kind of like well, we think containerization's
going to be explosive and, you know, really everything's on virtual machines or bare
metal; mostly virtual machines. So, one, you know, as containers come out, how do we track them, secure them? How do we even secure, you
know, the virtual machines in our environment because there are almost a quarter of a million of them. The idea of being able
to put network policy that's, I would say, incorruptible, not actually on the
server, but at, you know, that's why we use firewalls, right? So solving that security
problem was number one. The other one was being
able to have the telemetry to see what's happening, what's changing, and troubleshoot at, you
know, at the network layer from every single server. Again, it's all about scale. Like things were just scaling
and the through puts going up, traditional methods of being able to see what's on your network, you can't look in the middle, it just can't keep up. It's just speeds and feeds. So being able to push
those things to the edge. And then, lastly, it really happened more through the process here, but about a year and a half ago, we began segmenting our network the same way a 5G provider does, it's technology called segment routing, and we just said, that's kind
of our follow-on technology is to, you know, put the
network in the server and put the segment routing
capability all the way out at the edge. So, you know, some things
we foresaw and other things we've just developed, you know, it's been, it's been two and half years, so it's been a great partnership and, you know, I think
more features will come. >> Soni, you and the team, it's been talked about all day long, have a history of
multiple times that you've kind of brought these big
transformational technologies ahead. What did you guys see
a couple of years back and kind of this progression
that you saw this opportunity to do something a little bit different than you've done in the past,
which is actually go out, raise a round and do a real startup. What was the opportunity that
you saw this challenge here? >> So we saw a number of
challenges and opportunities at the same time. We clearly saw that
the cloud architectures that had been built by the leaders like, and incumbents like AWS, today have a lot of the
intelligence that is being pushed into their respective compute platforms. And we also noticed that, at the same time while that was what was needed to build the first
generation of the cloud, the New Age applications, and
even as Gartner has predicted, that 75% of all enterprise
data and applications will be processed at the edge by 2025. If that happens, then you
need that intelligence at the edge. You need the ability to go
do it where the action is, which is at the edge. And very consistently, we
found that the architectures, including scale out storage,
were also driving the need for this intelligence to be
on, in a scale out manner. So, if you're going to
scale out computing, you need the services
to be going hand in hand with that scale out compute architecture. For the enterprises, so they can simplify their architectures and
bring the cloud models that have only existed in the cloud world into their own data centers
and their own private clouds. So there were these technology transitions we saw were coming down the pike, it's easier said now in 2019,
it wasn't so simple in 2017 because we had to look at these multiple technology transitions and surprisingly, when
we call those things out as we were shaping the company's strategy, getting validation of the
use cases from customers like Josh, was pivotally
important, because it was further validating that
this would be the direction that the enterprises
and the cloud customers would be taking. So the reason you start with a vision, you start with looking at where
the technology transitions are going to be occurring, and getting the customers
that are looking further out, validate it, plays a very important role so that you can go and focus
on the biggest problems that you need to go and solve. >> Right. It just seems like the big
problem for most laymen is the old one, which is why networking
exists in the first place, which is do you bring
the data to the computer, do you bring the compute to the data? And now, as you said, in kind of this hyper distributed world, that's not really a viable
answer, either one, right? Because the two are blended
and have to be together so that you don't necessarily
have to move one to the other or the other back the other direction. And then, the second piece
that you talked about over and over in your
presentation was security. >> Yes. >> And, you know, everybody
talks about security all the time and everybody gets hacked every day and there's this constant
theme that security has to be baked in, you know, kind
of throughout the process as opposed to kind of
bolted on at the end. You guys took that approach from day one. >> Yes we did. >> Bake it into the architecture. >> Yes, that was crucially important because when you are
trying to address the needs of the enterprise, particularly in regulated
markets like financial services, you want to be in a position
where you have thought about it and baked it into the platform ground up. And so when we are building
the programmable processor, we had the opportunity to go
put the right elements on it in order to make it tamper proof. We had to go think about
encrypting all that traffic and communication between
our policy manager and the distributed services
platforms at the edge. We also then took it
a step further to say, now, if there were to be a
bad actor that were to attack from an operating system
vulnerability perspective, how do we ensure that we
can contain that bad actor as opposed to being propagated
over the infrastructure. So those elements are
things you can not bolt on. At design time are when
you need to go put those into the design, day one. Only on top of that
foundation then can you build a very secure set of services,
whether it's encryption, whether it's distributed
firewall services, so on and so forth. >> And Josh, I'm curious
on your take as we've seen kind of software defined
everything slowly take over as opposed to, you know, kind
of single purpose machines or single purpose appliances etc. Really, a different opportunity for you to control but also, to see, a lot of talk today
about policy management, a lot of talk about observability, and as you said, now even
segmentation of the networks like you segment the nodes and
you segment everything else. How do you see this kind of
software defined everything continuing to evolve and
what does it enable you to do that you can't do with
just a static device? >> Well I mean the approach we took, we started, like, you know,
years ago, about six years ago, was saying, we can get computers deployed for our applications, no problem, and, you know, on demand
and in our internal cloud, now we can do it as a
hybrid cloud solution. One of the biggest problems
we had in software-defined was, how do you put security
policy, firewall policy, with that compute? And then, you know, our industry, there's lots of segmentation for material non-public
information, compliance, you know, it could be internet
facing, B to B facing, we do that. Today we program various
firewall vendors automatically. We allow our application developers to create these policies and
push them through as code, and then program the firewall. What we were really looking
to do here is distribute that. So we, day one, in getting
Pensando into production was to use our firewall
system, it's called Pinnacle, we program from Pinnacle directly into the Pensando vennis manager via API and then it, you know,
uses its inventory systems to push those things out. So for us, software-defined
has been around, I like to call it the store
front, but for the developer, it's network policy, it's load balancing, and that's really what they see, those were the big products on the net, everything else is just
packet-forwarding for them. So we wanted, with Pensando, at least starting with security, to have that bar set, day one. And then get, you know,
all the benefits of scale, through put and having the
policies close to, on the edge, you know, we're back to
talking about the edge. We want it right there
with the deployment, with the workload or the application. And that's what we're
doing, right off the bat. >> Yeah. One of the things you
mentioned in your talk was kind of in the theme
of atomic computing. You want to get smaller and smaller units so that you can apply and redeploy based on wherever the
workload is in the chains. And you've said you've now
been able to, you know, basically take things out of
dedicated, you know, kind of a dedicated space, dedicated
line, a dedicated job, so that you can now put them
in a more virtualized situation and grab more resources as you need them. >> Well you would think
the architecture, I mean, even just theater of the
mind is just, you're saying I'm going to put this specific
thing that I have to secure behind these firewalls. So it's one one catenative
computers or a hundred, it's still behind a set of firewalls. It's a very north south,
you know, get in, get out. Here, you're talking about having that same level of security and
I think that's novel, right? There hasn't been, if you
look at virtual firewalls or, you know, IP tables on Linux, I mean, it's corruptible, it can
be attacked on the computer and once you've been attacked and that attack vector
has been, you know, hit, you're compromised. This is a separate
management plane, you know, separate control plane. The server doesn't see it. That security's provided,
it's at scale, it's east west. The more computers that
have the Pensando, you know, architecture inside of them,
you know, the wider you can go. And then the north south goes away. >> I'm just curious to
get your perspective as you know, everyone is a technology company at the same time, technology
budgets are going down, people are hard to hire, your data is growing exponentially, and everything's a securities threat. >> Yes. >> So as you get up in the
morning, get ready to come to work and you're drinking your coffee, I mean, how do you kind of communicate to make sure that senior management knows kind of what your objectives are and this kind of ongoing challenge to do more with less in IT even though it's an
increasingly strategic place, or if it actually is what
the company does does, it just happens to wrap it
around airplane services or financial services
or travel or whatever. >> I think you're, and I'd
said it to John before, it has to come, that budget
has to come from somewhere, so I think a combination
of, of one that's less, well, I'll say the one
that's easier to quantify is you're going to take budget from, say, appliance manufacturer and
move it to a distributed edge. And you're going to hopefully
save some money while you do it. You're going to do it at scale,
you're going to do it at, you know, high through put
and the security is the same. Or better. So that's one. That's one place to take capital from. The other one is to say,
can I use the next computer? Yes. Because I don't have to deploy
these other new computers behind this stack of firewalls. Is there agility there? Is there efficiency? Am I buying less servers
and using, you know, more of what I have and doing it, you know, able to deploy faster? And it's harder to quantify. I think if you could,
you know, over time see, I bought 20% less server capacity or, you know, X86 capacity, that's a savings. And the other one that's
very hard to quantify, but it's always nice to have
the development community and we've had it recently, where they say, "Hey, this took me a month to deploy instead of a year." And, you know, the
purchase cycles, you know, for procurement and deployment, they're long, you know, in enterprise. You want them to be quick
but they're really not. So all of those things add
up and that's the story you know, I would tell,
you know, any management. >> Right, I think, you
know, the old historic way, the utilization rates were
just so, so, so, so low between CPU and memory
and everything else. Because, if nothing else,
because to get another box you know could take like a long time. (laughs) >> Yeah, exactly. >> Well, final question for you Soni. You talked about architectures and being locked into architectures and you talked about you guys
are already looking forward you know, to kind of your
next rev, your next release, kind of your next step forwards. What, where do you see
kind of the direction, don't give away any secrets, but, you know, kind of
where are guys going? What are your priorities
now that you've launched, you've got a little bit
more money in the bank? >> Well our biggest priorities will be to focus on customer success. Is to make sure that the customer journey is indeed replicable at scale. Is to enable the partner success. So in addition to Goldman
Sachs, the ability to go and replicate it across
the federated markets whether it's global financial services, health care, federal. And partnering with HP
Enterprise, so that they can, on their platform amplify the
value of this architecture. Not just on their compute
platforms but in other areas. And the third one clearly
is for our cloud customers, is to make sure that
they are in a position to build a world-class cloud architecture on top of which then
they can build their own, deliver their own services,
their own secret sauces. So that they can excel
at whatever that cloud is whether it's to become
the leading edge platform as a service customer, or
whether it is to be the leading edge softwares
service platform customers. It's all about the execution
as you heard in that room. And that's fundamentally what
we're going to strive to be. Is to be a great execution
machine and keep our heads down and focus on making our customers and our partners very successful. >> Well Soni, congratulations again to you and the team on the launch today. And Josh, thank you for
hosting this terrific event, and being an early customer. >> Yeah, yeah, happy to be. >> All right. I'm Jeff, this is Soni,
Josh, we're at the top of Goldman Sachs at the Pensando, the new, welcome to the new edge. Thanks for watching. We'll see you next time. (upbeat tech music)
