|
|
Written on
Nov 29 2011
3,622 views,
|
New! Listen to this post in our daily podcast.
by
Brian Madden
It's Tuesday, and time for another Brian & Gabe LIVE, our live internet radio show where we talk about the latest news, trends, and technologies in the desktop virtualization and consumerization of IT industries. Each week we invite a guest to join us on the show. This week we're welcoming Chetan Venkatesh (@chetan_) Chetan is the founder and CTO of Atlantis Computing, as well as a two- (or three-?) time BriForum speaker.
Discussed on today’s show:
- Brian and Gabe Live is done in the cloud on the cheap—this is the real democratization and consumerization of IT and media.
- An overview of VDI—it’s an old joke that this year is going to be the year for widespread VDI, but Chetan sees more maturity in the industry and more defined roles for companies in the ecosystem as a catalyst for increase adoption.
- Layering and its role in VDI—hint: it’s more important for companies that are doing more advanced VDI deployments. Chetan called this VDI 2.0.
- User-installed-apps—Will it throw a wrench into VDI, or can it be dealt with?
- An overview of Atlantis, including some insight about how ILIO works, and the direction (management versus storage) that they are going in.
- Chetan announced Atlantis Storageless VDI, where the OS is in RAM.
- What will happen as the cost of IOPS goes away (like the cost of CPU did)? To solve storage issues, do we throw IOPS or optimization at a problem?
- Finally, we concluded by talking about some consumerization issues and how we came up with FUIT.
Tune in next time for more, and remember, you can chat with us to have your questions answered, live on the air.
Transcript
Brian: Hey, good
morning from San Francisco, on the 29th of November 2011. My name is Brian Madden. Joining me in the studio in San Francisco, as
always, Brother Jack.
Jack: Good
morning.
Brian: And our guest
today Mr. Chetan Venkatesh, CTO and founder of Atlantis Computing, and
three-time BriForum speaker I believe, and contributor to brianmadden.com, so Chetan,
thank you for joining.
Chetan: Hi, Brian.
Hi, Jack. This is a real pleasure. Thanks so much for having me on the show.
Brian: And always,
of course, Gabe is coming to us from Omaha.
Gabe: You bet, how
you doing Chetan?
Chetan: I’m
doing awesome. It’s just I’m sitting
here in your studio marveling at this amazing piece of technology, and
marveling at the fact that –
Brian: All that we
pulled off.
Chetan:
– 99 percent of this stuff is on the cloud, I mean wow, what an idea. And we’ll talk about that.
Brian: We talk now
about that you know, so Chetan walks in, and is like so what is this thing
exactly? And I said, yeah, it just
doesn’t matter, put on the headphones, talk into here, and you’re allowed to
say fuck if you want.
Chetan:
Oh, awesome.
Brian: And that’s
essentially, you know, that’s how it is.
But, yeah, this whole thing we were talking is cloud based. I know we mentioned that before, but we
actually use Amazon’s – they’ve got an offering called Cloud Front, so it’s
Amazon media encoder in the cloud, it’s Amazon edge network, they’re global,
they’re GO located DNS hosting, all of this kind of stuff, and this entire
show, I’m not kidding, costs I think are built Amazon. It’s like $4.00 an hour.
Chetan: Wow.
Brian: Because we
turn on the whole stack a half hour before the show. We built our credit card as soon as the
show’s over. We cut down the stack,
terminate the whole thing, and the clock stops ticking.
Chetan:
That’s amazing.
Brian: So, it’s – we
just about put the show together on our college credit card. You know like a $500.00 credit card. We’ve done a lot with very little to make
this show happen.
Chetan: In
many ways, everybody tends to think You Tube is this great democratization of
media, but to me in many ways this is taking it several levels, several notches
above that. I mean, go back ten
years. One bid, it costs to reach out to
say 500,000 people over a show like this.
Probably cost hundred thousands of dollars, and at $4.00 for an R, wow.
Brian: Yeah, now if
we have 500,000 visitors, it’ll probably up to a little bit higher, but –
Chetan: Yeah, well.
Brian: – maybe ten
bucks. But that’s, I mean yeah, it’s
crazy. And we do almost I think have
500,000 listeners, plus or minus a few half million or so.
Jack: Or 1,000.
Brian: Yeah, so
Chetan, I mean, as I was telling you earlier we began this show based on the
fact that Gabe and I were always talking to each other on the phone for like
and hour, and we’re, like, you know man, we should just record this because
people would like to hear what we say, I think.
Gabe: And you know
best part is we can’t hear them talking back and saying, oh, those guys don’t
know anything. Although we do have the
chat room.
Brian: Yeah, but we,
you know Chetan, you and I, I mean so first of all you drove up here from
Mountain View?
Chetan: Yeah, well Belmont San Mateo.
Brian: Well, thank
you for coming so early. It’s really
goddamn early to get up, and drive up here.
But you and I get together, I don’t know, every few months, and we have
dinner, whiskey, and cigarettes, and just kind of chat about the industry, so
we haven’t done that in a while, so let’s catch up, so how you doing?
Chetan: It’s
been great, Brian. It’s probably been
intense here for the industry as a whole.
There’s just so much going on.
Everybody has gotten happy, and then resigned and cynical about VDI, and
then sort of now you’re starting to see some elements of hope, and you know
anticipation that things are changing, things are maturing, and it’s finally
gonna be the decade of VDI or whatever.
Every
year is called the year of VDI, and that’s an old joke already, but I think
what we’re starting to see some light come through the clouds now finally. There’s, I think, a strong momentum in
adoption now finally around VDI. I for
one am seeing fairly large VDI installations, 5,000 to 15,000, 20,000 users
happen. And I predicted three years
back, what was it, 75 percent of all Windows desktops will be consumed through
some sort of a VDI, sort of a model.
You
know we’re not there yet, but it’s certainly is inching towards that at a
glazier pace, and it’ll happen, and that process just like a glazier carves an
entire geology around it. I think VDI
will carve a completely new eco system around this model of fatigue, and
certainly all that you’ve done with discussing consumerization of IT I think in
many ways is evident to that happening already.
There is just a complete disconnect between how enterprises do Windows,
and service delivery of desktops versus how users think their IT needs to come
to them. And that disconnect is just
driving users to be more innovative, more creative, and because they have
access to such new services, in such an easy way. A credit card literally, they’re reaching out
to them.
And all
of that is a part of this VDI momentum.
It’s not isolated, this whole movement away from the physical into a
data center, and into the cloud somewhere.
Brian: So what do
you think is the catalyst for that now, you know – if it is an old joke now,
that you know next year “will be the year of VDI”, but now you’re saying it’s
actually starting to happen. Is it that,
is the technology better? Has the price come
down, have people reset their expectations to be lower to where they can now actually
see VDI be successful?
Chetan: A
combination of all of the above. I
certainly think that there was a tremendous amount of gap in the technology
itself. Each lender was a part of that
ecosystem. In many ways it had not
figured out where they fit, and how to interlock with the others. I mean go back three years, and five years. Atlanta for one was everything to everyone,
and there were many like us, right. And
what’s happened over the last three years is there’s been a strong
crystallization of what each vendor, what each part of the eco system does very
well, and it interlocks with what is underneath it, and what is about it in a
much better way.
And
that as I think in many ways eliminated a lot of the technology risk that came
with trying to adopt VDI. I’ve got
customers who probably attempted VDI five years back, and 80 percent of the
effort that they put into doing this was their own. Vendors kind of, but their 20 percent. I think it’s flipped around now where you can
pretty much buy 80 percent of the stack today.
At 20 percent is your own effort in terms of customizing that, and
making it work for your organization.
The
other thing I think is I tend to talk about VDI in an enterprise context. I do not talk about it in so much in SMB or
consumer context because to me those are overlays that are happening around VDI.
But to me, the enterprise desktop, and that going into the cloud is
really their sense of VDI.
Brian: Do you think
that – you know I think that Atlanta’s computing was probably early on for me
one of the company’s that really started embodying this whole layering concept,
and now we have other companies doing it.
I mean we’ve got Virtual Computer, we’ve got Moka5, you know Ron
Oglesby’s chat room. I assume that’s
true, and so there’s a lot of people doing layering –
Chetan: Yeah.
Brian: – now, but I
mean is – and you guys are doing layering.
I mean is this idea that we’re gonna have sort of a layered dynamically
created desktop? Is that the model of
Windows and VDI or is it more about just moving the existing machines they have
into a data center delivering Windows in a 1-to-1 way?
Chetan: So, I think that the layerization, the de-componentization
of Windows from a technical perspective has been done, and there are some great
technologies out there. Unidesk, Wanova
certainly come to mind, and at a technical level, they like think figured out,
and covered the entire surface of the problem area. But from a Windows management deployment
delivery standpoint, the organizations that buy these services are not mature
enough to rethink their approaches.
There are customers who are, there are companies that are, and they tend
to go to this non-persistence stateless model straight away.
They
embrace that opportunity to change what they see as 15 years of problems,
whereas others have invested in those problems at such a fundamental level that
it’s very hard to throw the baby out with the bath water in many ways. That’s how they see the problem, and so they
tend to take these baby steps, and baby step No. 1 is let me just take the
physical estate the way it is. You know,
stick it in the data center actually the way it is, and then enter it over that
in small increments to see how it can get to statelessness, and that’s really
the bulk of people out there attempting VDI.
And
those are the people that I think in many ways have been cynical about it
because taking that approach is actually extremely complex. But if you take the de-componentization, the
layering approach, it’s actually much simpler, although there’s a much bigger
learning curve, and intellectual effort involved in that is far higher, but if
you’d attempted that and you did it, you actually get there much faster, so
it’s both of these aspects. There are
two sides of this VDI climb in many way.
Brian: It’s like, I
don’t know what the analogy is here, it’s like a surface tension thing. You’ve got higher service tension to go to the
sort of layered dynamically constructed approach, but once you break through
you can sort of sling-shot, like actually change the way that you deliver
Windows.
Chetan: Think about what you’ve done over here with
the cloud, and I’ll try to make this analogy.
It’s going to be fairly nebulous, so bear with me, but let’s go back ten
years, and the recording studio like this was, I don’t know, a million dollar
investment. And the ability to broadcast
this to even 100 users, 1000 users, was so high. In fact, the minimum, I guess, viable
population for this to make sense was I don’t know, maybe 100,000 people
listening to this show. If you add
anything less than that it wasn’t possible.
But you’ve had all these components sort of break apart, the studio’s
become virtual. All the pieces have sort
of flowed in and out, and what makes since to be in the cloud has flowed into
that, and as a result now, once you’re willing to put all these pieces
together, and rethink this entire thing, you’re gonna actually make your
minimum viable population as low as 100, and get away with it for $4.00.
Many
ways does de-componentization, and layerization works exactly the same
way. You’ve got to be willing to make
that investment, but the minute you do that it dramatically alters the entire
cost picture, the entire outreach, and the way you can actually do Windows and
desktops.
Brian: So I’ve
always been skeptical that the layering actually works because I felt like the
core technology to isolate applications over here, and user settings over here,
and operating system stuff there. That
was easy enough, but what happens when you get into scenario where the user
installs an application, and then to add almost a patch of layer below it, but
that lower layer is not compatible with that application. Doesn’t that throw up some kind of exception,
and then you have to disable some application if it doesn’t work right, and
don’t you have enough of those exceptions that it just breaks the whole think,
and layering just kind of doesn’t work for people?
Chetan: Well, it’s not a blanket approach for
sure. You’ve got to treat it like a
pyramid. I think there’s a big
population of users, and if it sort of drawn out as a pyramid that it works
for, and there is some population at the top that it doesn’t work for, and if
anything, there is a lot of improvement in UIA.
UIA is now compared to where it was even 24 –
Brian: UIA being
User Installed Applications?
Chetan: Yeah, exactly, User Installed Apps. It’s come a very long way, and there’s also a
commoditization happening over there. I
mean Absence is now giving that away as a part of their product portfolio
right. So it might now cover 100 percent
of the surface, but it covers end up off the surface for it to be a very viable
strategy for the vast majority of your users.
Not necessarily for everyone, and it doesn’t –
Brian: And nobody’s
doing it quite the same way.
Chetan: I’m
sorry.
Brian: Nobody’s
doing it quite the same way. The UIA,
it’s kind of like AppVirt, like everybody has a slightly method of doing it,
and so like you mentioned now absence is giving it away, but absence as a
solution maybe isn’t as robust as say unit desks or Wanova’s or Moka5’s.
Chetan: Sure,
you know, the beauty of my job is I get to talk about all these things just by
reading about them, so I won’t get into which ones better, and which ones not,
but –
Brian: Oh, no, no,
no, I’m not saying that at all. No, no,
I understand.
Chetan:
At a broader level every lender needs to differentiate, and so everybody tells
the same thing a little different way, but they’re fundamentally similar
approaches, but yeah, everybody does them differently, sure.
Brian: And so you
say you’re starting to see some decent VDI deployments, you know 5 to
15,000. Are these people doing layering
for some portion of those users or are they doing layering for all their VDI
users, but they’ve only brought over the used case that could be layered?
Chetan: Yeah,
so what typically tends to happen is layering is sort of a second version of
their VDI, so it’s VDI 2.0 for them. And
that’s where they’ve basically done the VDI 1.0. in many ways amortize the pain
of getting that up and running, and now are looking for value, and efficiencies
around the life cycle of the desktop. In
VDI 1.0, lot of the times it’s security, it’s capital investment, it’s
refresh. Those sorts of very capital
decisions that drive it, and once that is done efficiencies and life cycle and
operations are the next set of fruits that they want to capture, and so a lot
of customers I think tend to attack the layerization as a second stage. Once they have a fundamental understanding of
what it means to put this Window’s desktop in the data center. So I would say I think it’s more like 80/20,
80 percent of the customers are doing VDI 1.0, just fork lifting this thing
onto the data center, and then 20 percent of them are doing VDI 2.0 which is
now looking to reengineer the desktop, and capture those efficiencies.
Brian: So one of the
things I’m curious from everyone when I hear that they start to see sort of VDI
out in the wild. What do you see in
terms of both platforms, and just based on your knowledge, does it seem like a
lot of people using Hyper-V or XenServer and then what about are they view
customers in the XenDesktop, or the view workspace or they what?
Chetan: So
I think we used to see a lot of XenDesktop out there, and now we’re starting to
see a lot more View customers out there, so it’s started balancing out, and I
would say from my perspective, it’s almost 50/50 at this point.
Brian: Oh, really.
Chetan:
Yeah, yeah. It’s – and the new release
of View I think has got some great features as well that has helped that. From a hyper wise perspective, ESX have been
the dominant platforms. In the US we
still tend to see a lot of them just running on Vesure in their data centers
because that’s what they’ve instituted, but in Europe, we’re starting to see a
lot of Hyper-V. For whatever reason
there’s a lot more appetite for Hyper-V, base deployments over there starting
to see that. Xenserver is their licensing
just makes so much as sense that a lot of customers opt for that, and so I’d
say Hyper-V and Xenserver were sort of on par, and but EXS is still by far the
most dominant platform for the Hypervisor.
Brian: For these
people who are doing VDI 2.0, are these people who started VDI, I don’t know,
five years ago, and now they’re sort of reengineering their VDI environment, or
do you see people jumping straight to this sort of rethought desktop built on
layers where they’re doing that is their first jump into VDI?
Chetan: So, it’s usually – this is a long tail of
Window’s applications within that organization that dictates whether they do
2.0 or 1.0. The more apps you have
virtualized or available in a package format to deploy, the easier it is to go to
that 2.0 model. So customers with the
limited entry of apps, and it’s more in line of businesses, typically embrace
through auto straight away. They don’t
want to go through the one auto process, but customers who have a very large
entry of applications, who have a multi effort just virtualizing them, and
packaging them up, they’re the ones that do the one auto and the two-auto, so
that to me sort of seems to be the way this particular problem gets
attacked. It’s the invention of
applications and how do they packaged that up.
Brian: So if anyone
has any questions or wants to join this conversation, feel free to post that
into the chat. It’s Dan Brinkman, also
past BriForum speaker, and one of the members of the Denver Virtual Desktop
Drinking Association. Kind of rights to
all of this, you know, AppSense’s is not even released yet, and here we’re
saying oh, user installed apps, oh this is free now actually, but –
Chetan: It’s
free if you are already have AppSense.
Brian: A – I guess
that’s true to, so a lot for us. It’s – I wonder if there’s people I mean
because layering doesn’t necessarily mean user installed apps, so I guess, or
does it? I mean I guess –
Chetan: It
should.
Brian: – should.
Chetan: That’s my opinion. I think that once enabling technology is
there, there’s no reason not to use it.
Of course there are enterprises that will always have a philosophical
concern about that, and that’s a matter of time, but UIA I think is a very
important aspect of it. And if the
technology can support it in a way that alleviates these security concerns and
management concerns, then by all means go for it, use it, carpet bomb it.
Brian: In order for
the technology to support it from sort of a security standpoint is that
something that layering fixes or is that done by something else, and kind of
security software, policy software I’ve got running in my desktops.
Chetan: I
think that’s the big evolutionary piece that’s now happening within UIA, which
is people are realizing that it’s not just a filtered driver that redirects all
this crap, and puts it in some sort of a user layer. It’s once the policy engine that allows
things to come through, things not to come through that can enable/disable
etcetera, and that’s the next piece that’s sort of getting layers. I don’t know enough about each vendor’s prop
to tell you who has it, and who doesn’t, but that certainly seems to be now
sort of where most of these guys are innovating, putting that policy in layers
and rules around how these layers are created and consumed etcetera, and decommissioned,
all of that stuff.
Brian: So I want to
ask you about Atlanta’s computing.
Chetan:
Yeah.
Brian: But first
let’s take a quick sort of thirty-second commercial break because hey, we have
a sponsor again this week, so today’s episode of Brian and Gabe Live sponsored
by Citrix GoToManage. Citrix believes
that while desktop virtualization is loaded with benefits, it’s not gonna give
you a free pass on desktop management.
In fact, if you didn’t have a need for remote management and desktop
support before, you definitely do now in a virtual desktop environment.
So this
is where Citrix online can help. They
have a cloud-based IT server tool called Citrix GoToManage, combines best in
class remote support with network monitoring and alerting capabilities. Giving IT professionals like you total
control of your IT world. With Citrix
GoToManage, you can instantly deliver live support to customers or access
unattended computers and servers. You
can monitor the performance of desktops, servers, and networks.
You can
stay on top of issues with proactive alerts, and now there’s GoToManage Mobile
which provides support right from your iPad.
So if you want to try Citrix GoToManage for yourself, just
GoToManage.com, click on the free trial button, and enter the promo code
Madden45, and the Madden45 promo code, it’s a special code for Brian and Gabe
Live Centers which makes your trial last 45 days instead of whatever the lower
standard is, so desktop management, tech support, and monitoring has never been
so easy. That’s Citrix Online’s
GoToManage. Thank you so much, Citrix,
for your support.
Back to
you Chetan, and so I want to talk a little bit about Atlantis Computing
specifically and obviously you know this is – we always, I’ve always enjoyed
talking to you, but since you are the founder of Atlantis, let’s just get an
update because I think I first wrote about Atlantis in 2009 or something like
that, and I remember at the time, well anyway, I wrote about Atlantis back in
the day, we’ve interviewed Jim Moyles, an Atlantis employee also, another
BriForum speaker. We’ve gotten updates
from him but, just kind of give us 30 seconds on Atlantis, and then update us
on sort of where you guys are now.
Chetan: Sure,
it’s been – I started the company in ‘07, so it’s been four years, it’s been a
great ride. We’ve in many ways helped
build this ecosystem, we’re an informed part of it, and as a founder it’s very
satisfying to now see crystallization of so many of these things that we
envisioned four years back happening.
From a company perspective, we’ve actually been more or less out of the
startup mode, and do more of a real business.
We’ve added tons of new customers in the last 12 to 18 months, 60 odd
new customers – a lot of them being repeat customers, so from a business
perspective the business is growing as VDI sort of just trickles upward.
We’ve
been focused a lot on the storage side of the equation even though initially we
started off dealing with the decompentization and the layers that could be used
to manage Windows better. But
fundamentally, what we had was a piece of technology that at a storage level
was content overt. This is a piece of
software that speaks NTFS, and speaks storage and virtualization, and so we
decided that really the best place for us was to add value to the hypervisors
by bringing content awareness to it. And
so we built this virtual machine ILIO that basically optimizes VDI IO and
eliminates all the challenges that come with storage in a VDI deployment.
And
that has worked really well for us; we’ve built a very strong business around
that with 70 odd partners. And some of
the largest deployments globally, I know this one will probably get some eyes
rolling because we always talk about it, but JPMorgan Chase – I think 45,000
desktops today in production, and that’s probably the biggest one from a
layered perspective. This is actual
deployment of layering, managing, using layers, etc. – and 45,000 global users
on that platform, so that’s been an enormous success for us. But outside of that, a little more mainstream
I would say success with VMs, with CB Richard Ellis. In Europe they’ve standardized on us, got
some very large federal customers in defense, etc.
And
we’ve also moved towards to some new areas of innovation so we’re actually
going to announce, and I’m sort of going to preempt the marketing guys by
announcing it because I – this is the show to do it at.
Brian: Hey.
Chetan: We’re actually announcing something that we
called storage less VDI in a week. And
what storage less VDI does is it actually eliminates the bulk of storage that
is required to power VDI for running the operating system in the application
working set. And it does that by using a
piece of RAM memory, ultra-fast memory right, as the storage layer for running
the operating system and the application.
And just the user data gets shunted away to some sort of persistent
storage. So if you look at VDI, the vast
majority of storage that’s being built into VDI is just there to run the
operating system and run the applications.
It’s not really there to store data, it’s there just to store the VMD can,
power the IOPs that are required to run Windows.
And so
we’re actually delivering that off just RAM and we’ve highly integrated this
with Cisco and, so the Cisco UCS platform is what we’re making this offering
on. They’ve got some amazing technology that’ll
let you do high memory density blades, and so we’ve got to blade the B230, for
example, with 512 gigs of RAM that you put an ILIO on and you support 160
desktops. There’s a half a million IOPs
within each blade, and so needless to say, IOPs is no longer your problem.
Brian: I guess
not. Is money my problem?
Chetan: No,
you’d be set right, and it’s a great question.
The actual cost per desktop is about $200. It’s actually cheaper than any form of
conventional storage that you threw at this thing. So it’s a great piece of innovation because
it addresses the cost angle, it addresses the infrastructure angle, and it’s
trying to solve the problem in the right way.
And we’ve got about three customer deployments for this already so –
Brian: On UCS?
Chetan:
On UCS, yeah.
Brian: Is that –
does your three customers using UCS encompass all customers who are using UCS?
Chetan:
No, I don’t think so. I think it’s a small population. But if I look at the UCS, I think it’s the
ultimate VDI machine. Lots of memory and
lots of CPU, and the ability to have the back channel just move IO very quickly
between blades. So it is in many – VDI
is the killer app for this machine.
Brian: So I want to
ask you more about that, but Ron’s asking in the chat, he says the ILIO IO
stuff rocks. Are you still doing top of
the rack appliances and VM appliances?
Chetan: That’s
a great question Ron and shout out to you, thanks for that question. So we actually are just a VM appliance, we’re
nothing but a virtual machine. And by
the way, we also have support for now an Xenserver and Hyper-V. We got to run all three hypervisors
today. So that’s actually another piece
that I want to get out there. But, and
it’s a VM appliance, and as a result it can actually be deployed on top of a
hypervisor, either top of rack or underneath.
All the changes is the kind of storage adaptor that you put through it,
and the number of VMs that you would house on each machine. If you’re top of rack, it’s great for a
conventional persistent desktop model.
You simply stick on top of rack, plug it between your storage and the
VDI rack and simply migrate V Motion, your users from their existing VDI estate
or from PtoV them from physical virtual, and you’re off and running.
It’ll
automatically get optimized and it’ll run off this thing. But if you truly want to leverage
statelessness and all the lovely layering stuff that’s out there from unit desk
and all these guys, you can actually deploy us on each server. And what happens when you do that is
essentially you can leverage local storage on each host, for some of those layers,
it sort of tiers the storage automatically for you, but more importantly, all
the IO stays within the host, and will go as fast as the CPUs will let
you. So you essentially get ten gig or
more within that host without actually even investing in that ten gig.
Brian: Is that all
in memory then or do you need to have like a fusion IO and SSD or something
like that?
Chetan: It can actually leverage anything, so it can
leverage pure memory as in the case of UCS and storage lists, and VDI.
Brian: And the pure
memory is what will be – like do you need UCS for pure memory or can I – if I
have just regular rack mount servers, can I put an ILIO VM on each host, and
then have it, that VM have memory that’s used as my storage?
Chetan: Well,
we’ve got some integration points with UCS today that leverage the high memory
densities, so once those sorts of integration points are available on other
conventional commodities servers, yes.
But today the memory solution is fairly UCS specific. But you can stick an SSD in there, you can
stick infusion IO in there, hell you can even stick in a bunch of SAS drives,
and rig them together, and use that as your local storage department, and this
is going to optimize all IO –
Brian: Wouldn’t I
have to fill this room with SAS drives to get a 500,000 IFs?
Chetan: –
well, here’s what we’ve got, a certified configuration with four 15K SAS drives
that gives you 68,000 IOPs on an ILIO.
So if you want 68,000 IOPs which is more than plenty for the 60, 70
desktops that you’ll run on –
Brian: Well, I guess
ILIO itself is reducing the amount of IOPs you need per desktop anyway because
you’re getting sort of that – does it still us that flock thing that kind of
filed –
Chetan: –
yeah.
Brian: – block
level.
Chetan:
So what’s happened is that flock stuff went from being more about a management
interface to being an underlying IO and NTFS processing technology. So what we’re doing is we’re sitting in line
fronting, and creating fake virtual machines that the hypervisor boots and
runs. And as that IO goes back and
forth, we’re actually using this characterization that is the flock to, I got
to be really careful how I say that which is –
Brian: No you don’t.
Chetan: We’re
using this flock technology to basically characterize, figure out what that IO
is. Is that a paging request, is that a
change to some specific user file or is that a registry edit? And all of those are treated differently by
the ILIO clients, so it basically gets characterized, figure out what it is,
what the request is, write, read, and what particular NTFS level object. And then second level de duplicate that,
remove it off the wire. And that’s a
pretty neat trick because we don’t actually commit the data anywhere before we
move to the wire. It’s actually just
picking it off.
Brian: So it’s not
like a batch process, its data de-duplication?
It’s never duplicated in the first place.
Chetan: Exactly, it’s complete single instancing of
the write side, and the read side as a result of that. And then it comes into a special intelligent
caching mechanism which is very NTFS specific, and so different objects get
cached in different ways, and serviced different ways. And ultimately it gets written to disk in a
very efficient way.
Brian: So how do you
know how to do all this?
Chetan:
Well –
Brian: This is – I
mean this is some hard core stuff.
Chetan: – it is, it is some crazy stuff.
Brian: Some dude
sitting here like where’d you learn this stuff?
Chetan: Well, I’d love to take full credit for it and
I will, but I honestly I would just complete embellishment of – we lucked
out. We got some really great engineers
who had worked in conventional storage systems, and were frustrated with the
way these storage technologies were trying to fulfill VDI. So we lucked out in that we got some really
hard core guys come aboard with understanding of file systems and layouts,
NTFS, etc. And so ultimately it is was
just a combination of the right talent, with the right ideas, and some kick ass
engineering management that sort of built this product for us.
Brian: So Atlantis,
I remember when we were talking to you, two and a half years ago or whatever,
it was really about that whole layering and ILIO was about managements.
Chetan: Yeah.
Brian: And then you
just sort of, I guess part of that became, the way you did the management, you
could really handle just super-fast IO especially across a lot of similar type
machines, and I know that it seems like Atlantis went to focus on the server
space, just accelerating virtual server storage for virtual service too.
Chetan:
Right.
Brian: So how, do
you feel like Atlantis now is the focus to accelerate storage that works for
servers and desktops or is there a desktop management play still in here? Or is that even a bullshit question to say
hey, what are you man, desktop management or storage?
Chetan:
We still have the desktop management stuff and it’s in deployment. We have several customers who use it, but
it’s not the front and center of our business.
We’re a storage optimization company, and we support all management
layering products. And personally, my
own interest is in the plumbing, and how to make shit go faster. That’s really what I care about, and those I
think are very interesting, compelling innovations to bring to market. The layering stuff is important, and I think
there’s a lot of innovation in different places. In some ways I wouldn’t call it better, but
probably attacks the surface in a much cleaner way.
So our
layering makes a lot of sense when you’re a certain style of deployment, and
you’re willing to think the ILIO way, but if you’re more of a conventional
shop, the product, there are more generic options available, that’s how I put
it. So we’re really storage
optimization, we’re really about changing the form and function of the
storage. We’re about accelerating
desktops, we’re about accelerating servers.
We’re about shrinking the storage footprint. And where it makes sense, we have a layering,
and a management piece that can help you provision better, clone faster, and
manage those desktops in a more efficient way.
Brian: So it seems
if you look at the world of storage as it applies to desktops and VDI, there’s
sort of two, I don’t know what I want to call them, waves coming together
right? One wave, I could say it’s
companies like Atlantis, like UniDesk, Wanova, Moka5, all the layering type,
management type companies, and they’re working on single instance, whether it’s
file blocked combination, whatever. But
then there’s also the more traditional storage companies. I mean I’ve looked at Extreme IO, look at
Nutanix to some extent, a lot of these startups who are looking purely at the
sort of storage appliance level. Maybe
Atlantis kind of rise both of those, but these are coming together.
Do you
see consolidation? Like is this
something where the management bore into companies are all bought by bigger
storage companies or just smaller kind of single instance block-level companies
are bought by bigger companies and in two or three years is all this just, you
buy storage from HP or Dell or EMC, whatever and you just get all this
functionality anyway? Or do you think
this stays as its own niche kind of add on?
Chetan: I
think there’s going to be consolidation, it just has to happen. So it probably plays out in a much bigger
time frame, it doesn’t play out as simply as X will buy Y. But one thing for sure, I think everybody’s
starting to appreciate that decompentization is a very important part of layer
management. And who the layer management
vendors are, is fairly nebulous. Is it
Microsoft with their way of approaching it or is it Citrix for example with the
RingCube acquisition right? So there’s
definitely going to be consolidation, and conventional storage companies are
very challenged right now to find a strategy, and a product and a technology
approach to solve this problem and they will buy where they can.
There
is all sorts of interesting discussions right now of even non-conventional,
non-necessarily storage or VDI companies trying to enter this space. There are companies that are –
Brian: Like?
Chetan: –
well there are companies that for example are in the robo market, and they’re
thinking about how do you deliver VDI to the remote office in some
non-conventional ways, rather than just accelerating the network? The ICEA or RDP, but how do you actually
accelerate the storage function itself, so there’s all these interesting
mechanics that are in movement right now.
Brian: And I guess
when you wrote a few years ago that you thought whatever it was two thirds,
three quarters of future instances of Windows will be delivered in some kind of
VDI like model.
Chetan: Yeah.
Brian: At the time I
thought that was hog wash, most of it will be local. I’ve since turned 180 degrees, and now I’m
100 percent on board with you – we can talk about that in a minute.
Chetan: Sure.
Brian: But then Tim
Mangan kind of called us out, and said no, Brian and Chetan are both wrong,
it’s really about data and where the data is.
And I guess data and storage are almost the same thing. Well, we talk about storage; it also enables
where the Windows runs and that kind of stuff.
If you look at the data, there is an angle there also right? Because if you look at it like do I really
need my desktop or do I just need my Dropbox contents, and so maybe that’s what
you talk about these other angles. So I
guess a company like, well maybe it’s like a Dropbox or comp provider or look
at the network acceleration vendors who look at these that they might take this
approach, and flip it completely where they’re looking at solving the same
problem that VDI solves, but not with centralized copies of Microsoft Windows,
but rather making the data, so you have whatever you need, wherever you
are. And I guess that’s kind of where
you’re going with this when you talk about how these different vendors could
come together.
Chetan:
Yeah, absolutely. I think that in some
ways it’s a redundant point of view.
Data was the first to move to the cloud in the first place before all
this shit went in the cloud, data was there already. I mean you started using Dropbox before, and
it was a much easier way to get to use and consume that service then it is to
put Windows in a data center, so data just – where it’s stored is
immaterial. The whole conversation of
whether it’s local or whether it’s in the cloud is a stupid conversation because
it doesn’t matter as long as you can get access to that.
Brian: So I’m like
writing the article, Fuck the SAN and your articles Fuck the Conversation about
Fuck the SAN.
Chetan: Exactly, so
in many ways it’s just that. So what
data there in the cloud or wherever you need it to be, really the heart of the
question is where do you run Windows if it is not strategic, if it’s a tactical
tool for just running applications.
What’s the most efficient way, and we’ve covered this before right? And the answer is, it’s on an easy tool like
cloud where you get these large economies of scale. And you can drive more users per server, per
rack and drop that cost down and make that minimal, viable population right in
– as small as possible, so it has to lead that way because that’s what makes
fundamentally cost sense and efficiency sense.
Brian: So based on
that, people who aren’t doing sort of VDI today have there reasons for going to
VDI or has the barrier to VDI significantly changed? You said that more people were doing it, but
I mean if I’ve got a bunch of traditional desktops today, do I wait another
couple of years and wait for all this.
Now you’ve got with UCS only in memory based local server kind of using
it over there, but if I wait another six months or year I can have the on all
platforms and if I wait another year I mean.
Do I want to wait a couple years, and do it once we fix the storage
problem or do I want to wait just myself right now?
Chetan: So
I’ll take a little bit of a different angle on this one, Brian. So I think that what VDI is tied into is tied
into a much bigger macro set of trends.
It’s not tied into a technical reason to do it; it’s tied into some very
economic reasons. The No. 1 being that
there is a drive towards efficiency in every enterprise in America. Everyone wants to get as efficient as
possible. And a part of that is they’re
restructuring the businesses to basically outsource as much as possible. So almost every customer I talk to has that
as a very hot issue that might be latent within the organization, but it’s
certainly driving things. So what
they’re trying to do is build like a desktop fabric, a compute fabric, that can
service their conventional businesses as well as that new model of doing
business with outsource providers etc.
It’s a
very fluid environment. And they look at
VDI as fundamentally being that fabric that can be extended out into the
outsource and back in the in-source and within the corporation etcetera, so
that tends to drive a lot of this conversation.
It’s not just that the technology pieces are there. And so in many ways the amount of energy
that’s underneath the VDI movement is much higher than we appreciate as
technical folks because we tend to look at check boxes, and see if this stupid
thing works with that stupid thing, and say it does and I need you to happen
now. But underneath it are just much
bigger macroeconomic forces at work.
And
fundamentally the world economically is changing. It’s an amazing time to be alive and watch all
this, if you can get past the drama. But
the reshaping is very fundamentally tied to how this compute fabric is
delivered to people who are part of organization, wherever they are, and VDI is
the best, most efficient way to do that.
Brian: A bunch of
folks in the chat are – there’s a common theme that they’re talking about.
Chetan: It’s
like a IOPs competition.
Brian: Yeah, and
well, IOPs per dollar, per Euro, whatever you want to call it, so all of this
stuff, do we – does everything we’re talking about specifically run the
technology for storage? Does the fact
that IOPs get cheaper and cheaper and faster and faster negate the fact that we
need any of this or is it truly is there kind of saying we’re going this way
anyway, and in a year or two we won’t be thinking about do we do storage sort
of the old way or the new way? There’s
only just the way it is, and the concept that we have just blocks duplicated
any wherever, we just seem like a sort of archaic quaint thing from yester
year? But my question is does the cost –
does driving down the price of IOPs and everything just mean that all of this
we’re talking about, all of these companies reshaping the way storage works in
two years is cheap enough that we don’t even care. I don’t have to mess with it, I’m gonna be
able to buy 500,000 IOPs in my blade and I won’t need any of this?
Chetan: No,
it’s, I think IOPs is one very big cost component right? And the cheaper that IOPs, that component
gets ,the better it is from an overall stand point. But it is still just one component. It is not the only component, there is still
compute. Compute is already dirt cheap,
CPU cycles are essentially free, and there’s memory. And memory and storage are sort of blurring
in many ways. They are very amorphous at
this point, and I think truly at some point there is no distinction between
what is memory, and what is persistent storage.
But IOPs – you will get to zero on IOPs, it’s already happening, and
when you do get to zero what’s interesting it’s not just VDI that happens, but
interesting new applications that were not possible before because you were
IOPs constrain that what happened before.
I’m
going to use one of the favorite theories that I have from George Gilder who’s
one of my favorite writers, and Gilder said that every new innovation takes
advantage of some surplus right? To take
something that was in a very limited form and make that a surplus right? And so if you look at, for example, what CPUs
did, they essentially allowed you to create CPU cycles, and each generation of
Moore’s law gives you more CPU cycles, so you got to surplus the CPU cycle, and
at some point you got a technology like the hyper visor that then efficiently
harnesses all those CPU cycles, and gives you a much more higher form of
value. So we keep going in these
incremental waves, and so when the IOPs challenge is fulfilled, when the cost
of IOP goes almost zero, you’re going to see new applications evolve that were
not possible before, and I can't even –
Brian: I was fixing say,
can you tell us what those would be because I want that. Look at, we just signed a four year deal with
Tech Target, will I see that in the four years or do I have to do another deal
after that in 2016?
Chetan:
– well, you know I think there’s a lot of good research out there. I’ve been spending a lot of time reading on a
lot of fundamental research papers and you can see that that’s happening. There are a lot of smart kids out there, a
lot of people who are veterans of the software industry who have done their
exits, and they’re sort of going back and questioning what’s the next
thing. And one of the thinking points is
that the cost of IOPs goes to zero what next?
So it's definitely coming out there.
Gabe: Isn’t there
still a desire then for the optimization piece.
I mean it sounds like the same argument that we use for the network and
the plan where okay, so I’m being constrained somewhere. Either I can optimize it or I can throw more
bandwidth at it is. In this case we’re
talking about throwing more IOPs at where you might think bandwidth are, but
also I mean I’m saying this kind of keeping in my Dan Brickman’s thing in the
chat room – what?
Brian: I’m sorry,
you’re breaking up. So I think I told
everyone last week, by the way, that we do have a 100 meg connection coming in
here to the office, but that has – I think that’s this week maybe?
Chetan: Yeah.
Brian: So let’s take
a quick break and hopefully Gabe will be – your connection will magically fix
itself in the next 30 seconds, but I just want to take another commercial
break.
Gabe: – I’m gonna
call in.
Brian: All right, so
Gabe you call in, and while Gabe is calling in, I’ll say that this episode of
Brian and Gabe Live is sponsored by Citrix GoToManage, and I’m gonna paraphrase
what I said last time during our last commercial break, which is look, they
believe, and I can tell you personally I believe this too that when you go to
desktop virtualization regardless of if it's VDI client VMs, whatever, it’s not
a free pass to desktop management, you still have to its environment. So you need some kind of desktop management
for doing remote access, softer updates, fixing things, inventory software, hardware,
all that kind of stuff.
So
Citrix online has GoToManage which is essentially their cloud based desktop
management platform, and it has what you’d expect it to have, remote control,
inventory and all that kind of stuff as I said.
And they’ve got – so it’s a cloud-based service, you know everything –
their dashboard, all this stuff is powered by them. They’ve got a free demo you can try and you
can get the demo from GoToManage.com. If
you enter the code Madden45 then that would give you the 45 day trial for
unlimited computers in your organization.
I believe it’s unlimited. We haven’t
had to limit ourselves yet.
So
anyway that’s device management and tech support monitoring. They’ve got an iPad app that’s from Citrix
online at GoToManage.com. So again,
thank you Citrix for that. I’m not sure
if Gabe has connected back in yet. Are
you here Gabe?
Gabe: Yeah, I’m
back. Can you hear me?
Brian: Oh, you’re
telephone, that’s awesome. So anyway
tell – say again –
Gabe: Yeah.
Brian: – I wanna
wrap this conversation up in a couple of minutes, so we can get on to some
other things of the week, but share the thoughts – the question you were just asking
Chetan before the break.
Gabe: I’m not sure
where I dropped off, but my original or where I started was kind of in response
to Dan Brinkman’s message in the chat room about he feels like the Atlantis
stuff is really awesome, but he just can't shake the feeling that increasing
the number of IOPs or the increasing number of cheaper IOPs as time goes by
means that the ILIO product is, and products like that are gonna have a limited
lifetime. And my thought there and my
question to Chetan really is, I feel like this is kind of like the same kind of
conversation we had with Gwen about bandwidth where you’ve got two choices when
you have a constraint.
You can
either throw more bandwidth at it or you can optimize it, and so obviously
bandwidth is cheaper, so that’s no problem we can do that, but at some point in
time you can still, there’s still a lot of benefits to optimizing it. The same thing we’re seeing with storage, the
IOPs are getting cheaper and cheaper and cheaper, and we can provide more of
them in smaller packages and so on, but at the same time, it's still not
optimized unless we throw something like ILIO in there. So I mean is that your train of though, I
mean or do you feel like maybe I guess –
Chetan: Yeah,
so –
Gabe: – do you
think that ILIO has a limited lifespan?
Chetan: I
think that’s a great question, and it's – I’ll answer it this way, I think the
storage optimization and the IOPs was Act 1 for Atlantis, and Act 2 is to
unleash better management, the ability to do all sorts of interesting flow
control, service management, all of that stuff at the IOPs level, and
fundamentally allow you to get the cost of an IOP down to zero, and then move
up the stack to allow you to do better things with those IOPs. So I can't disclose more than that at this
point, but that’s one thing. And in VDI,
you know IOPs is a big deal, but there are also lots of server workloads that
have an IOPs challenge, and their IOPs problems are different from the VDI IOPs
challenge.
And
storage vendors will shake their heads and say, no there not, but it is. There are different workloads, and there is
no one size fits all and the beauty of software in something like ILIO is that
it’s algorithmic, and therefore you change the algorithmic to fit the workload. And so a lot of what we’re doing now is going
after service use cases while. For
example, we’re gonna have ILIO for XenApp product out that’s actually in beta
right now with a couple of customers.
Jack: Oh wow,
because Provisioning Server is really, really creaky right now. Like it’s getting up in age, and hasn’t see
too much development it doesn’t look like, so this can sort of reinvigorate
what Provisioning Server did for XenApp literally five years ago.
Chetan: Absolutely,
and everybody wants to virtualize XenApp or a lot of people do, and they’re
looking for – and the interesting thing is when you start doing that, it starts
to be able in some ways like VDI, and so there is an overlap and it becomes an
applicable technology, so we’re doing that.
We’re also doing some other server use cases that are a little earlier
on, so I won’t comment on those because we might change our mind. But it’s certainly an area that we’re going
after, and it is a challenge. Commoditization
is just a known factor. The hypervisor
window is commoditized when at zero, and they’ve moved out into other things,
and they’re trying to extract value and money from those things. We’re gonna do the same exact thing.
Brian: Okay, so
yeah, you definitely answered the guys question how do you make money when this
becomes zeros? So you just figure – that
–
Gabe: You don’t, you
do something else.
Chetan: Yeah, it’s cannibalization, it's all of –
it's a good problem to have let me put it that way.
Brian: – so in our
last twelve minutes here, let’s kind of shift gears, and get away from talking
about VDI and IOPs, and user installed apps and that kind of stuff. Gabe, talk about the series of articles that
you just started on consumerize IT. It’s
our new FUIT series.
Gabe: So this is –
when we started the talk about having a consumerization web site it was what,
maybe May, somewhere in there, and the whole time we’re like man, we’ve got to
go FUIT.com because this whole consumerization thing is about saying fuck you
to IT, and like so we have to go get this, and then we gotta have – maybe we
make it – maybe we have two sites, maybe we have one site that is like articles
and blog posting stuff, and we have FUIT.com that is all about screwing the
man. How to actually say FU to IT and
that kind of thing, and eventually we never really did get the go ahead to do
that, and so what we decided to do is when we launched a consumerization web
site, we would just have a little corner of it for the FUIT topic, and so
that’s what we did.
We kind
of started this, it's just a tag in our system.
We can go to a tag landing page, and see everything that has this FUIT
topic, and in there we’re gonna keep a list of all the different ways that
people are circumventing IT policy. And
so my, the very first post in there is an introduction like I just said, and
the second post is about how people can use email or can get around their
company’s corporate email restriction policy, people who we know, but may or
may not be us.
And in
those situations, the article that is up there specifically deals with a low, a
very small mailbox size, a low attachment size and a very short retention
period, and how all you have to do to circumvent that is to just set up a rule
to direct all your email off to a Gmail account where you have tons of storage,
and a 25 meg attachment, and it will retain that for as long as you want. And so, but we’ve done it in kind of a snarky
way with a fake company, and we made an enemy called the Powers, and so the
Powers will be our common enemy throughout all these FUIT posts that we do
ongoing. I’m shooting for one a week.
Brian: And yeah,
it’s funny. Where oh where did we get
the ideas for having to get around an IT system? So because the idea is – because this is the
whole thing on consumerzation anyway because a lot of people – as I started
talking about consumerization, and saying that users can kind of do whatever
they want, the audience that I talk to mostly is IT pros and desktop admins and
people like, and there’s always some percentage in the audience who’s like ,no
not happening, we’re in control, and I’m just thinking man if you think it’s
not happening, that’s – you just don’t know.
Like I
promise you it’s not even – and then we find also the people like in our case,
this is where this whole FUIT thing comes from because they say well, you know,
like at our company we have a SSL-VPN, but you have to have the right
antivirus, and all this kind of stuff just to get your file shares and your
file share and then you know what’s our solution? So we just use Dropbox, and they block
Dropbox so you just type a proxy in Dropbox that we run from home. That’s –
Gabe: And we talked
about it before too where you – when you joined Tech Target, and went to a San
Francisco office, nobody even knew what HTTP over RPC or whatever the Outlook –
using Outlook, pure Outlook without connecting the VPN. And nobody knew what that was, but we used it
at the old company, and so you – we figured out how to set our Outlook up to
connect via that, and then all of a sudden everybody – now nobody uses the VPN
to connect to their email. And so all it
takes is one person to do this, and then everybody does it. Once somebody discovers that all you have to
do is drive around the security gate, then everybody is gonna drive around the
security gate.
Brian: – that’s a
picture of Jim. That’s you
colleague. Jim posted a picture of a
little gate across a road and, but it was in the snow, so you could see all the
tire tracks in the snow driving around the gate cause it was literally just a
gate and a road, but the field next to the road had no fence.
Gabe: Yeah, it’s
similar to the one that we use in presentations. I think we’re gonna have to switch it though
cause Jim’s is way better.
Brian: Yeah.
Chetan: This is
where I get excited about a product like Horizon because I think in many ways
what it's gonna do is give the suits a friendly face, and at least satisfy some
of the user population that today is saying FUIT. If you can put an App store inside every corporation,
at every enterprise, and allow a syndication process for Apps to be brought in
by the IT organization. And I think IT
organizations, fundamentally just, you know it’s the typical set of human
reactions. The first human reaction
always anger, the second one is denial, the third one is depression, the fourth
one is acceptance, and the fifth one is let’s do something about it right.
Every
organization, every IT organization does exactly the same thing, and so when
your blindsided, and the first thing to do is to get angry about it, and force
it in a heavy handed way, then to get depressed about it, then deny it,
etcetera. They do the same exact things,
but ultimately the constructive way to solve that is to get behind the users,
otherwise your just gonna fail that job.
And so I think Horizon and that whole set of tools, Citrix Dazzle for
example, I don't know what happened to that.
Brian: It's now
built in – what the functionality of what Dazzle was is built into the Citrix
receiver platform, so it’s just called receiver now.
Chetan: It’s
fantastic because that is gonna allow that syndication process, and that
provisioning process to get behind enabling users, so if you want drop box, you
want some service out on the cloud, you’re gonna get it, but you’re gonna get
it in a way that makes everyone comfortable.
So I don’t necessarily –
Gabe: Yeah, I don't
think people –
Chetan:
– Sorry go ahead, Gabe.
Gabe: – sorry, I
don’t think that people are out there trying to maliciously get around IT
policies or business policies or whatever those restrictions are there in
place. I don't think people are –
there’s not that many people that are at least actively trying to circumvent
them, it’s just when they are presented with the problem they just do what it
takes to get their job done.
Brian: Yeah.
Chetan: Absolutely.
Gabe: Yeah, I think
you’re right, I think if IT embraces it I don't think – I think that will go
along way for making everybody happy.
Chetan: Yeah.
Jack: And I think
that’s an example, even if there are a lot of people trying to circumvent what
IT puts in place is not because they want to do it for the sake of doing it,
it's just like if they’re given the opportunity to have something else then off
they would go. And so with this site or
with this FUIT column specifically, we had the initial idea of like it’d be fun
to make how to guides like, how to configure Outlook RPC over HTTP. How to use drop box instead of corporate
VPN. How to set up a proxy to use AIM at
work.
Chetan: They’re like the pirate bay of –
Brian: Yeah, but
then we realized if we did that man, we’re gonna get every single like
15-year-old and like 60-year-old who finds it via Google, and then it’s
emailing us like oh, I don’t see it because the Outlook Service Pack 2, they
moved the checkbox over here, so we decided to flip it around and write it more
– which is more honest and work appropriate for us anyway to make it more for
IT pros to read, which is oh, by the way, here’s how your users are doing this
kind of stuff.
And I
always tell the story at these events when I was visiting a fairly large, like
100,000 plus person company, and I was talking to the CIO, and she was talking
about how they have all this like great policy where they block attachments
from non-secured devices, so it’s like that’s how they get around
security. And then I walk out of that
room with my PR person, and the first thing she does, the PR person working at
this company, and the first thing she does when I was in the room is pick up
her iPhone, and say yeah, yeah, yeah, no send me the presentation. No, no, I’m on my iPhone, send it to my
Gmail.
Chetan: Right.
Brian: And I want to
tell the CIO, like I don't want to screw the CIO, but I want to say, so you
know that – and what are you gonna do block Gmail? I mean, come on.
So
they’re gonna – like the sender has Gmail too and if you block Gmail then
they’re gonna send it from their MiFi, their 3G connection, so it’s just one of
those things. So Chetan, I don't know if
you have – you’ve built yourself a nice little real company. How do you get around your internal IT stuff,
or do you get like ultimate control since you’re CCO?
Chetan: You
know it’s funny, it’s a different part of the lifecycle ,and I see my company
in some ways becoming that type of a company that you always make fun of. So we’ve standardized on a laptop model, and if you don’t want that laptop model you
buy your own, but we support that thing a lot.
We’ve been cloud friendly from day one, so almost everything that we
have is out on the cloud. And it’s not
gotten as crazy and heavy handed, but I certainly see people just using whatever
tool makes sense for the job.
If
Dropbox makes sense, you know people screamed about the Dropbox agreement in
saying they think they own all the content or something like that, and there
was a lot of palpitations about that, so we made certain accommodations. We don’t put some of our intellectual
property stuff on Dropbox for example.
If you’re writing up a patent or something like that, you don’t put that
on there, you put that somewhere else, so.
Brian: You hope they
don’t put that there. You ask them
nicely not to put that there.
Chetan: Yeah, you can and –
Brian: I bet you
patents are in Dropbox right now.
Chetan:
– you’re just basically, ultimately relying on the good sense of people to do
what’s right, and you hopefully can educate them. We’ve got – we don’t have a full-time IT guy
yet. We don’t need to because in many
ways we’re –
Brian: Yeah, good
size company for no full-time IT person.
Chetan: –
we’ve got a part-time guy, and it's just because that we sort of said use the
right tool for the job, and try and educate people about what the dangers
are. That’s about the extent I think you
can do. It's like parenting man, I mean
ultimately if your kids gonna go and smoke weed, he’s gonna go smoke weed. All that you can do is educate them, and give
them the best of everything, and hope that they make the right decision, so.
Brian: If we give
them the best pot –
Chetan: Well you
could tell –
Brian: – good
parents don’t let their kids smoke ditch weed.
Chetan:
– I don't know, you know it’s basically like you have to give up that illusion
of control.
Brian: Well, I tell a
story. We had a family in our
neighborhood growing up where the parents allowed their high school age kids
drink in their own basement, and they took away the keys, and they kind of said
the same thing like hey, they’re gonna drink anyway, so I’d rather have them
drinking and puking in my own basement instead of driving around running away
from people which is a very controversial thing, but all those kids who are down
there are still alive today.
Chetan:
Right.
Brian: And I hate to
be so crass on that, but that was it. So
in our last minute or so Chetan, I want to ask you, I want to revisit something
that we talked about earlier that a couple of folks in the chat who just asked
if you can quantify the 50/50 between the Xendesktop and VM View. And so I think –
Gabe: I think they
want us to quantify that.
Brian: – well, yeah
so –
Gabe: Since we’re
independent.
Brian: – well, I
mean is it – so what your saying 50/50 View versus desktop is this – what your
proofs of concept, new deployments because I think people are surprised that
View has such a large market share from what you’re saying, again just what
you’re seeing personally.
Chetan: So
most of our production deployments are driven by Xendesktop simply because of
the presence it had before. A lot of the
POCs now are view based, so 50/50 is more reflective of some of the
initiatives, some of the new POCs etcetera.
What we tend to see with larger deployments are they’re Xendesktop based
because of that form architecture, etcetera allows you to do whatever. And I think View will sort of catch up, and
get there as you start to see these rollouts.
But 50/50 really is a POC perspective, not production deployment. A lot of production is still on Xendesktop,
yeah.
Brian: All right, so
that’s actually the end of our hour. It
actually went by pretty fast. So we have
–
Chetan: Yeah.
Brian: – there’s a
lot more to talk about, but luckily next Tuesday is only seven days away, so
next week we’ll be back same time. We
have a – we’ll kind of pick up this conversation where we left off. I know the chat people are talking about User
Installed Apps. I think we should
probably bring on a guest like Rueben Sprout in the future and talk about
that. We should probably bring on
Claudia and talk about BYOC. Ryan
Oglesby, if you’re still out there, we should bring on you and maybe talk about
UYA some more. And I think for all of
you who are listening today, thank you so much for taking the time to Citrix
Online and there GoToManage which you can download trial for 45 days with a
Madden 45 code. Thank you Citrix for
sponsoring. Chetan, thank you so much
for making the super early drive up here.
Chetan: Hey,
no problem, it’s a pleasure Brian and Gabe and Jack. Thank you so much for having me here.
Brian: And Gabe, thank
you for dialing in. Jack –
Gabe: Yeah,
hopefully next week we’ll have it fixed.
Brian: – yeah,
hopefully we’ll have a faster connection next week. Jack, thanks for coming to early, you didn’t
have to come in as far.
Jack: Of course.
Brian: And again
thanks for listening. So from San
Francisco this is Brian Madden, I will see you next week.
(Note: You must be logged in to post a comment.)
If you log in and nothing happens, delete your cookies from BrianMadden.com and try again. Sorry about that, but we had to make a one-time change to the cookie path when we migrated web servers.