1 Comment


At Structure, a panel of IT execs and investors discussed the enterprise IT winners and losers of the past few years. Here’s what they had to say.

Transcription details:
Input sound file:
1006.Batch 4

Transcription results:
Session Name: Useful or Useless? The Technology Maturity Level Decides

Paul Santinelli
Arne Josefsberg
Peter Krey
Raj Patel

Announcer 00:05
Do not leave. Instead, actually, get out your smartphones, get ready to either call your broker or get ready to process a mobile trade, because we have a distinguished panel who is going to answer the question: useful or useless? I think that the answer is probably yes, but they’ll probably have more depth to that. Without further ado, Paul Santinelli, North Bridge. Welcome.

Paul Santinelli 00:35
Thank you. Far from distinguished, let me just clarify that [chuckles]. So, we come to these conferences all year long, and for many of us, many years have been showing up to Structure and other conferences, and we listen about all these great technologies that everybody’s pitching and talking about, “It’s the next thing! It’s the next generation of X!” But the question that I came up with was, “How many people are actually using this shit [laughter]? Is there actually somebody using the technology? Is it actually making a difference in the world we live in?” I’m not looking to change the world. I think what we’re looking to do is change the face of technology. So, without further ado, we’ve done a ton of prep on this, we wrote a long script, and then my panelists and I rewrote it about ten minutes ago [chuckles]. So, Arne Josefsburg, the CTO of ServiceNow. Peter Krey of Krey Associates, also known as ‘The Hired Gun’ for many of the technology banks on Wall Street. And Raj Patel, who happens to be at Cisco. What we’ve done is, we’ve come up with a list of what we think are the winners and the losers based on things that have been spoken about and talked about for years and years and years, and where we are today. Arne, I’m going to give you the first one.
Paul Santinelli 02:09
Thank you.
Paul Santinelli 02:09
On the winners side.
Paul Santinelli 02:10
On the winners side?
Paul Santinelli 02:12
On the winners side we talk about SAS adoption.
Paul Santinelli 02:15
Obviously, being at a SAS company for the last little less than two years, I’ve actually been surprised about the rapid adoption of SAS. We were talking about SAS for years as an industry, there was a lot of talk. I actually think it’s happening at fairly massive scale right now, at least that’s what I’m seeing. I think SAS is a winner.
Paul Santinelli 02:42
We still have the enterprise resistance for SAS, so Raj, you may want to talk about some of the things that you’re seeing, both in your previous role and potentially in a future role of yours.
Raj Patel 02:57
Yeah, so I manage the Cloud for WebEx, which is real time online meetings and conferencing, and one of the things that Arne and I were talking about is SAS and carrying the bag, in terms of delivering it everyday, is actually how you engage with enterprises and this little thing called software and change control. So, many of the speakers here before talked about how you deal with the enterprise. It’s the trade off between agility and control; agility and control from a change standpoint. You have to work with the risk proof tolerance profile of your customers and figure out what deployment model works for them; whether they need to be on a dedicated software cluster so that they can kind of move at the pace of their organization, or if you need to move 100% SAS and move at the speed that the Cloud provides. So, those are some of the interesting nuances that I think, while somewhat in a soft world it’s pretty dramatic when you actually get out there.
Paul Santinelli 04:05
So, let’s talk about our first loser, which is near and dear to my heart. I’m going to have a completely contrary point of view, but Arne, you and I were talking about this. You felt that SDN has kind of gone nowhere.
Paul Santinelli 04:18
Yeah. Actually, some years ago, I was on the board of something called the Open Networking Foundation, which was sort of an early implementation of SDN, and I’ve got to say the conversations of today three, four years later sound almost exactly like the conversations we had four years ago. So, I don’t think it’s moved very far. I’m also thinking we’re over-optimistic about how is this going to be adopted by enterprises? It’s one thing to do it in Facebook or Google, but doing it in enterprise is a totally different thing.
Paul Santinelli 04:54
The other thing I kind of picked up today that kind of makes me a little worried is, it seems like each implementation, each vendor of SDN is sort of going down their own path. That, I think, is very, very problematic. If you really want to serve the industry, there should be a common SDN implementation.
Paul Santinelli 05:13
27 controllers and nothing to control.
Paul Santinelli 05:15
Exactly [laughter].
Paul Santinelli 05:18
That’s essentially what we’ve seen with the majority of the SDN vendors, except I’ll give the gratuitous plug because I happen to be an investor and on the board of one that actually has real customers doing layer four through seven technology, which is Embrane, which has taken a little bit of a different path, which is, ” Let the guys try to control the control plane, we’re going to focus on the other layers of the network and stay out of the mix”.
Paul Santinelli 05:43
So, let’s go to the next winner. SSD, near and dear to your heart, Peter.
Peter Krey 05:49
So, I’m a big fan. I think that SSD technologies are hot to trot and coming. Densities are increasing, dollars per unit are getting much better, and I’m also a big fan of one of the start-ups out on the East Coast, not well known out here out West, called Whip Tail. They’re one of the few out of the whole industry that can actually do very large scale-out versus single device scale up, multi-million up raise, etc, etc. They’re doing some phenomenally interesting software layer enabling of SSD’s versus the limitations I have of a 400 MHz ARM chip inside of an SSD device.
Paul Santinelli 06:26
Raj, we’re going to pin you down on this one because you brought it up as a loser, and considering you’re with the largest networking vendor in the world, IPv6.
Raj Patel 06:36
Let me qualify that [chuckles].
Paul Santinelli 06:37
How many people want him to qualify that and just explain why he said it was a loser [laughter]?
Raj Patel 06:44
So, let me explain why I think it’s a dud. I used to be the guy – I was at Yahoo before Cisco, and obviously at WebEx in the SAS world. I think again, to roughly what Arne is talking about is, just as an industry participant, relative to the wolf cry, or whatever, the “world’s going to stop” thing that IPv6 was over the last couple of years, generally the world’s kind of gotten along – at least from my perspective – and I’m a networking guy also, at heart. When I calibrate how much emphasis I used to place on it relative to what it turns out, that’s when it kind of gets to a dud state for me.
Paul Santinelli 07:28
So, you like the old security vendors that wrote viruses to sell more security software [laughter]? Got you. Good job. Let’s talk about one of the winners: Cloud to enterprise integration. We’ve all seen what’s going on with AWS and I think a lot of enterprises are having a difficult time on two vectors: the cost vector, which is, “Hey, at a certain point it no longer makes sense for me to use eight of you. It’s more costly for me to roll my own.” The other vector is all the pieces that come with using an external service; security, roll your own, what version of software, what’s my stack look like?
Peter Krey 08:13
So, let me throw in a few comments on that. I’m probably not the expert, but we work with enterprise customers and large enterprises have three, four, five thousand applications behind their firewall. I hear a lot of people talking about, “Oh, they’re going to move everything over to the Cloud very quickly.” The reality is, you’re not going to move 3, 000 applications to the Cloud anytime soon. We’re talking about 10, 15, maybe 20 years. So, this notion of, “There’s going to be compute and storage and bandwith behind the firewall for a very, very long time” is a reality. So, I think the players who are going to figure out how to bridge the public Cloud with what’s behind the firewall are going to be the winners here.
Paul Santinelli 08:56
I think we saw one of those today. If people haven’t seen it, you should stop off and try to find the Meta Cloud guys. They’ve done a real impressive job of sort of duplicating what everybody wants to duplicate in the enterprise and they’ve put together a pretty interesting solution. I think Raj, you know those guys pretty well and you can comment on their direction.
Raj Patel 09:15
When I think about winners, OpenStack seems to be that. If you’re really at a macro level, if you think about something that has the potential of having the impact that I think we all hope for in the infrastructure business, it does, and if we assume that – and I agree that moving a significant amount of workloads to OpenStack is still to be done, but then if you look at companies that are doing interesting things to enable that technology, Meta Cloud and a couple others that come to mind that are taking that platform and then enabling enterprises to adopt it, which is really where a lot of the energy should be.
Peter Krey 10:04
Just to add into that, I also think using the Cloud to manage the things in your own data center, I think are very strong trends. So, you don’t have to move everything to the Cloud, but leverage the Cloud to manage things behind your firewall, as well as your Cloud resources.
Paul Santinelli 10:19
I think the contrary view to OpenStack is, “Are there too many chefs in the kitchen?” Are there too many independent vendors that are going to muddy the waters around OpenStack? I like OpenStack as a foundation, I love what those guys have been able to do. I do think it’s a winner conceptually. What concerns me is when companies like Red Hat decide that they’re going to ship their own OpenStack alternative. That kind of breaks away from what we’re trying to do in building the standard.
Paul Santinelli 10:50
If I can chime in on that one. One of my colleagues here that’s a visitor at the conference, we’ve been looking at OpenStack for about three years. The original charter and the original design behind the thing was open source inner-operability. The really killer part of it is enabled Hybrid Cloud, so I could have a piece of code I ran on my internal cloud and I could move it to my external, and it’s completely compatible; no changes. What I’m starting to sense is a flashback to the 90′s when I used to write applications for Unix. So, if I wrote it on AIX, it wouldn’t run on HPUX, nor would it run on Solaris. So, we’re a little skeptical, we’re a little cautionary here to make sure that when we write the app it actually can be portable and you don’t have to touch it and everything really is inner-operable, but that remains to be proven.
Paul Santinelli 11:35
So, let’s talk about one of the next losers: carrier to cloud [chuckles]. I mean, this reminds me of the data center business, and if we’ve seen the data center business mature over a period of time, there were companies that were data center companies back in the day. It was Exodus, it was Global Center, it was Globix, right? Those companies are now the Dupont Fabros’ of the world. When I listen to carriers talk about being cloud providers, the only thing I can think of is somebody with a pair of spikes climbing up a telephone pole to turn the phone on and it concerns me. I just think that that’s a losing proposition. I don’t think you ever want to kind of walk into an environment where they control the pipe, they control the data center, they’re trying to tell you how to build your software stack, and at the same time, you’re sitting back looking at it saying, “Wait a minute. These guys couldn’t manage to do basic IP to the home ten years ago, we’re now going to let them run our enterprise apps”.
Paul Santinelli 12:40
It’s provocative [laughter].
Peter Krey 12:48
I’m with you [laughter].
Paul Santinelli 12:52
Chime in if you’re with me because I’m going to get a tomato from somebody from Verizon here in a minute [laughter]. How many people actually think that’s true? Would you trust your carrier provider to provide your application stack in the cloud? Come on, don’t be afraid. The Verizon people aren’t shutting off your service or dumping your data over to the NSA, I promise [laughter]. Actually, I don’t promise [chuckles].
Paul Santinelli 13:15
Well, that person sneezed, that doesn’t count.
Paul Santinelli 13:16
Let’s talk a little bit about the next winner. Peter, you brought this one up. 100% self-service IT, 0% IT in the organization.
Peter Krey 13:32
So, my favorite wisecrack – I have a slide that says, “Cloud means not enterprise IT,” and the moral of the story there is, going back to the 90′s, if you wanted to order a server or get some storage and get something built, you’re talking six to nine months of just process and governance, etc. The thing that really lit me up about Cloud is that, it’s 100% automated, and it’s 100% self-service, so what our customers and our colleagues can now do in three to six minutes used to take six to nine months, so to me, this is a gigantic game changer. It also enables a lot of folks to do things that are great ideas, they’re potential disrupters, and it’s a very, very powerful thing, so I’m very gung-ho about not being anti-enterprise IT – that’s where I was born and raised, but the problem is, we can’t have nine months of process when you can do it in nine months of automation.
Raj Patel 14:26
I think you had an interesting way of– in the earlier dialogue we talked about – and I think you asserted that – enterprise IT, in my view, is not actually going away, and the opportunity is to actually transition to an almost 100% service model. All of us who are in the “tech ops” world, or “dev ops” world have gone through that because the scale of the business. One of my architects articulated really well that we exist to create an agile application development platform for developers. That’s kind of how I would say the up-scale team kind of thinks about, and I believe you’ve kind of seen a lot of the DNA here. I think if that same mindset were to transfer in enterprise IT and be able to be self-service and be able to roughly get out of the way and just kind of think about it in that sense, I think there’s a great opportunity for relevance for enterprise IT and to be able to do things behind the curtains that make a lot of this stuff come together. One of the things that a lot of folks ask me, “Well, do they go away because everything just moves to cloud?” I think the gentleman from Amazon was talking about that folks are moving everything to the cloud – they could, but try moving a really high-end video conferencing system with FBGA’s and Asix to an Amazon or a public cloud; you couldn’t just yet. Those are the type of things where I think enterprise IT will continue to provide significant value, but they need to go to a transformation to a real service organization.
Peter Krey 16:00
That’s actually all we do is help IT in doing that. I spent several years in IT at Microsoft, and I think being in IT is one of the toughest jobs in the world. You’re never thanked for what you do; it’s a very tough job. I think is an incredible opportunity – this move to the Cloud – for IT to kind of get back into a leadership role. I think the CIO’s that sort of grab this and take it as an opportunity and sort of create this self-service consumerized experience, they’re going to be the winner. I heard someone earlier today talk about average for SEO is about 18 to 24 months; there’s a reason for that. It’s because IT is viewed as slow and it takes 18 months to get something provisioned; this is, for me, transformational right now for IT.
Paul Santinelli 16:57
I was invited to a dinner at one of these very large, global, farm of life science companies, and the table behind us were a bunch of the research PhD’s inventing different medicines and curing techniques, and I was at the IT table, and I could hear all the guys in back of me talking about them being able to run certain computational chemistry and other experimental things on a Saturday from home. They were able to explore a whole bunch of different medicine scenarios in a weekend; whereas, they couldn’t even get the meeting organized with their central IT folks to even talk about this. So, I’m not being anti-IT, I’m just being pro-automation and having everybody reallocated to doing customer service and taking care of the folks we’re supposed to help, not just be very processed and governance oriented.
Peter Krey 17:46
The interesting thing is, we live in the 60 mile radius cone of silence in the Valley. So, when you hear of Cloud, everybody must be doing cloud, right? You hear about IT and self-service, everybody must being doing self-service. It’s completely counter to the people that you work with on Wall Street. In most part, a lot of the IT professionals on Wall Street, they’re compensated like a banker. They have assets under management, they’re compensated by assets under management, they’re not compensated by efficiency, performance, flexibility, agility, so when you go to them and you tell them, “Hey, we’re going to build this self-service organization that somebody clicks a button and boom, they’re up and running,” they’re saying, “Wait a minute. That’s one less server in my organization, that affects my level and pay grade, and not my assets under management have gone down.”
Paul Santinelli 18:34
That’s a really good point. There are some organizational cultures that are still entrepreneurial, and of those, the adoption rate of this stuff is, orders of magnitude bigger than the traditional corporations. So, what you’re talking about are the traditional public companies and all the management models and compensation models are span of control. So, you manage 10, 000 people, I manage 1, 000 people, you’ll make eight to ten times more than I do. Whereas, in the old days, if you delivered something that enable a business to do something they couldn’t do, or you increased processing efficiency by one or two orders of magnitude, if you were impactful, that was a whole different ballgame. That’s what I think is really cool about the cloud automation; it may be bringing some of that back. So, it’s not just about how much we’re responsible for, it’s about how good a job you do and how impactful your technology is.
Paul Santinelli 19:25
So, let’s talk about something a little different: defining your stack. We know about all the people building tools around how to deploy, deployment tools, ways that you roll new code, ways you roll new software. Winner or loser? Puppet versus Chef. Let’s throw a bigger monkey wrench into that: ansible.
Raj Patel 19:55
Personally, we use Puppet in our stuff. I think it’s worked for us. I think we’ve also had some healthy debates from one versus the other, but that’s where I kind of come out a little bit. I would tell you this, one of the winners that I’m really delighted about, really the mass scaling of this notion of automated development to deployment to delivery to verification; that’s all we talk about. I think it’s become real. That, I think, is a phenomenally transformative things, and companies like Chef and Puppet have actually enabled some of that.
Peter Krey 20:38
You’ve taken an easy out.
Raj Patel 20:39
I’m not taking an easy out, I said Puppet [chuckles].
Peter Krey 20:41
I’ve met with both of those companies. I think it’s amazing, enabling technology. I think there’s room for two players. I’m not sure about the third, but definitely two players, and I think these two companies got something really good going for them.
Paul Santinelli 20:55
We looked at both. We kind of came up with a, “Tastes great, less filling; it’s a floor wax; it’s a dessert topping.” They’re both really good tools, but the thing is, who’s going to use it? In the Chef case, it’s much more procedural and structured, so people that have more a developer kind of style and bias, they’re way more into the ruby scripting Chef thing. The guys that are traditional systems admins, they like the Puppet 4GL kind of expressive language, and it’s actually really cool as long as it behaves in a consistent non-4GL way. So, depending on who you talk to, the world’s pretty evenly split 50/50 Chef versus Puppet. For us, when we used Chef, it is enabled automation levels. It’s just absolutely phenomenal. It’s a great tool.
Paul Santinelli 21:52
We have a few minutes left and we heard from Jay Parikh today on Facebook’s plans and what they’re really focused on and I think there’s probably a lot of questions about the fabric and how they’re looking at networking, and I think this group probably has a few views on that. Does it make sense for the Facebook guys – and Open Compute was clearly a necessity for them – they had to have servers that ran on per square foot, per watt that was much more denser capacity and much more efficient than anything that server providers can provide. Now, we’re hearing about a new switching fabric, new protocols that they’re going to develop in-house. How much sense does that make?
Paul Santinelli 22:37
I’m really good friends with Najim, I’m not quite sure if he’s here.
Paul Santinelli 22:40
He’s up in the top.
Paul Santinelli 22:43
He’s going to hate me for saying this, but I really don’t understand why Facebook would build their own switches. I understand their control plane. I think their control plane should be standardized and opened up for industry, but the switching side, I don’t get it.
Paul Santinelli 23:00
Raj, we have a minute 40 left, and I haven’t actually said anything about Cisco.
Raj Patel 23:08
That’s remarkable for you [laughter].
Paul Santinelli 23:09
It is remarkable [laughter]. Clearly, as one of the providers of this technology of switching and routing technology, what’s your view on what Jay and Najim are doing?
Raj Patel 23:18
Jay and Najim are good friends and we’ve talked about things. Understand, my role at Cisco is actually to be a SAS provider. So, in some ways, I kind of things about the world as one supplier or a technology provider. So, I’m going to provide more of a view as a SAS provider and as a network–
Paul Santinelli 23:41
You’re still paid by the Bear?
Raj Patel 23:42
Yeah, I’m still paid by the Bear, but the point being, I think for us, in the SAS world, you just look for core versus context. I kind of come out with, that is not my core that’s providing scalable world-class SAS applications and not a network fabric. Again, our challenges are probably a little bit different from a scale standpoint and things like that, but you can do, I would say, a non over-subscription network, a class network, and many of those things using, I would say, approaches that are now available in the world. So, I’ll probably find out from Najim over a beer why he’s thinking that, but whatever.
Paul Santinelli 24:24
So, we have 20 seconds. We’re going to give one person in the audience to throw out one technology and ask the panel, winner or loser. Who wants to go? Quickly. No takers? I got one in the front – Calxeta.
Peter Krey 24:42
No opinion.
Paul Santinelli 24:44
We’re testing, we’re bench marking, we’re looking forward to not only 32-bit server chips, but 64 as well.
Raj Patel 24:50
No opinion for me.
Paul Santinelli 24:51
Well, that’s it. We’re out of time. Thanks for weathering through this. We appreciate it. See you guys soon.

Announcer 25:00
All right. Couple of quick things before you all head out. First of all, I want to wrap up some of the lessons from today just so there’s some clarity around the first day of learning. According to David Giambruno, the Revlon CIO, private cloud is the way to go. According to Werner Vogels, who needs no introduction, public cloud is the way to go. According to John Engates, hybrid cloud is the way to go [laughter]. That’s the kind of clarity we strive for here at GigaOM, so hopefully now that resolves any questions that you may have. We also learned that [inaudible] can be used as dessert topping, I was completely unaware of that.
Announcer 25:40
So, we’re going to start tomorrow at 8:25 am, because 8:30 am would be too easy and too obvious. So, we’re going to do five minutes earlier. Please remember to bring your badges, although the GigaOM folks would like for you not to remember that because you’d have to buy another ticket at $1, 000 or $2, 000, so your choice, obviously. We’d like to thank our sponsors, Intel, the headline sponsor, and some of the others.
Announcer 26:08
There was one other thing left on the agenda, what… darn. I can’t imagine what it might be. Oh, that’s right, it must be open bar thanks to BTI Systems [laughter]. I will see you guys down there. See you soon. 8:25 tomorrow. Thanks.


firstpage of 2
  1. michaelbushong Thursday, June 20, 2013

    Part of the reason SDN is taking a beating is that we talk about it as if it is a tangible thing. It is a framework more than a solution. It’s a way of doing things. It’s a tool.

    The only people who buy SDN are CTO offices. And that is because their objective is to consume and evaluate new technologies. Everyone else deploys networks or application delivery infrastructures or cloud or whatever.

    We’ve actually put SDN in a tough position. It is important, but it is important in support of something else.

    Of course, when people ask, of course our products do SDN ;)

    Mike Bushong (@mbushong)


Comments have been disabled for this post