This document is a transcription of an interview of David Scott, SVP of HP Storage, on SiliconAngle.tv, at the HP Discover 2011 conference in early June. Interviewers were Wikibon.org CEO David Vellante and SiliconAngle.com CEo John Furrier.
DV: We’re here with david Scott, VP of HP storage and a Cube alum. Welcome, David.
DS: Nice to be here.
DV: So last time was at SNW and before that was Barcelona. So it’s been eight months since you’ve been on here.
DS: Eight months….
Contents |
3Par Integration into HP
DV: So you have to tell us: What have you learned, what are some of the surprises, pleasant and unpleasant, that you’ve found out at HP?
DS: It’s a much bigger company than it was when I left 10 years ago, that’s for certain. 320,000 people. $130 billion company. But it’s been a really exciting reintegration for me personally. A lot of people who I knew before at my career here at HP and it’s been really satisfying to see how well the integration of the 3PAR team into HP is going.
DV: So talk about that. Where are you at with 3PAR? Are you on plan, ahead of plan – you wouldn’t tell me if you were behind plan?
DS: We’re ahead of plan. And if you look at integration there are really three elements of this – people, the processes, and the integration of the products themselves. And we’ve been doing really well as far as the integration is concerned. I’m pleased to say almost all of our people have continued with HP; we had no significant attrition. That’s important to keep our engineering innovation going strongly. We’ve seen from a process perspective really smooth integration of our sales flow, support flow within the organization, moving manufacturing and our supply chain further and further into HP. And then from a product integration perspective we really have worked extremely fast. We’ve integrated assets like the Ibrix NAS capability with 3PAR, we’ve tied it into Storage Essentials which is the heterogeneous storage management capability; we’ve tied it into Cloud Systems Matrix, which is the management of the overall IT stack. And we’ve also got HP Two Gen Right Services certifying 3PAR’s utility storage platform as part of their outsourced storage services.
DV: So I think you’re saying that the engineering team, the people., and the organization sort of very receptive to the 3Par acquisition. We had Dave Donatelli on earlier and he kind of hinted that the portfolio in ESSN was not best-of-breed two years ago he joined HP, and one of the areas that needed attention was storage. You were relying on a lot of external storage and other suppliers, and 3PAR filled a big hole. So you’re saying the reception of the people inside HP was very positive. It wasn’t one of “What’s that mean for me?”
DS: No. I was delighted because in integrations such as these you can get a lot of antibodies coming out, and I can absolutely, categorically say that no such antibodies appeared. There was a huge amount of welcoming into the HP storage organization and in HP overall. The sales force was extremely excited; we integrated out sales team into HP very well. Out of the gate our first quarter we doubled our growth rates relative to the year-ago quarter when 3PAR was stand-alone. And we accelerated that growth rate into the second quarter. So the momentum has been very good.
The Future is in the Clouds
DV: You’re a real visionary. Before we dig deep into the business, I’d love to hear your take – you always have a good angle on what’s happening in the industry, the big shifts. What are you seeing now; what’s your telescope saying?
DS: I think the doubt about whether cloud is going to become a significant part of the future of computing, that debate has almost ended now. I think it is absolutely clear that, you know, ten years ago we might have been looking at a 30-year transition to cloud computing and people weren’t quite sure whether that transition would ever take place. But now I think people realize we’re 10 years in, probably 20 years to go, we’re going to go through phases of private cloud deployments, but everything’s going to eventually end up being the delivery of IT as a service. And I think that HP’s strategy with a connected cloud world is really in alignment with the way the world is about to evolve.
DV: I have a follow-up- on that. We just did a survey on Wikibon of our audience, about a couple of hundred people. And we asked them where they were at with cloud adoption, and one thing that stood out is that less than 10% said they were actively pursuing a hybrid cloud strategy. Now some of that could be definitional, but it struck us that maybe there’s a lot of marketing that’s ahead of the actual delivery. What do you see in terms of true hybrid cloud?
DS: I think it’s definitely occurring, and some of the most interesting hybrid clouds are actually areas where customers have become sophisticated enough that they are even moving to what I think is the next stage of cloud deployment, where they have not gone out to just one cloud service vendor but to two and are using those two vendors as dual operating entities and looking to provide DR between them as well as integrating with what remains of their core IT capability. I think that’s a very exciting vision of where things are going because a lot of people are most concerned about if they lock themselves into an IaaS vendor how do they ever get out. But if you start off with a couple of them, you always have that alternative. You can shop around for the best alternative.
JF: Multivendor cloud.
DS: Multivendor cloud, right. I think that’s where you will start seeing the importance of things like bursting capabilities being enabled, not just between private and public clouds but between the public clouds themselves.
Economics of Cloud
DV: So what’s your take on the economics of cloud. Everybody’s saying you can save a lot of money clearly, but there’s sort of the commodity cloud – Amazon Web services is the poster child – but then you’ve got the enterprise players sort of licking their chops at cloud. An example is VMware pushing the homogeneity of VMware everywhere, and it’s not clear that economically that’s the best way to go. What’s your take generically on the enterprise players’ ability to push the economics of their so-called private clouds into that hybrid cloud environment? Is there a discontinuity there?
DS: Well, I think the desire to have a single, homogeneous stack for your cloud solution is a misplaced view of the way cloud is. Cloud is going to have to reflect the breadth of the application environments that enterprises have today. And that means that you have to look at cloud solutions that in fact are open and can support multiple hypervisors. If you are a cloud service provider you probably need to provide VMware-based hypervisor support, Citrix end-server hypervisor support, and Microsoft HyperV support, because that reflects the diversity of application sets that you’re likely to find in an enterprise going forward. And I think the idea that it can end up being a single stack really is flawed thinking. It’s going to be an open, heterogeneous environment. And in fact HP’s cloud systems strategy is designed to address that very need. Rather than solely supporting one type of virtualization environment, it’s open and supports all of them.
DV: Does that mean more efficient, cost-effective stovepipes? Or do they all come together somewhere in the future?
DS: No, I think they all come together in terms of the fact that the underlying services around security, manageability, orchestration, provisioning, etc., are absolutely common. It’s just the hypervisor flexibility and some of the particular services that are available, and it may be that the integration of those hypervisors’ frameworks within this overall orchestration framework are available to customers.
JF: David, you’re fresh blood in HP after a very successful startup, and you’re in the front lines when you’re a startup -- you have to be nimble, resourceful. Now you’re in a big company. A lot of people want to know what its like in HP but also HP is doing and what is in the connectivity, networking, and the apps. Paul McCartney is going to be headlining here, so the question a la a little of Paul McCartney is: Is the vision for HP “Let it Be” or “Live and Let Die”?
DS: HP is a great pragmatic organization. “Live and Let Die” is not really reflective of what customers want you to do to them. They are in an environment where they have to transition their traditional IT environments to this new cloud world, whether its private or public cloud.
JF: This is really big that’s going on with cloud. You have to go for it. Baby steps?
DS: But fundamentally for most enterprises it’s going to be baby steps. It’s going to be a question of how do you provide customers the capability of handling this hybrid delivery model they’ll have – the traditional ….
JF: Maybe the Beetles song “Help!”
Services
JF: But on the services side of the innovation around cloud, where do you see the services changing? We talked with Donatelli about the innovation in services, where the delivery changes -- IT-as-a-service, cloud’s a big part of that and storage. You guys changed the game there. What do you see going on in the ecosystem around services, whether it’s consultancies out there or services that need to be delivered in this journey to the private cloud, so people can get that kind of help. People are looking for help. And david [donatelli] mentioned that the best product is the services angle.
DS: I absolutely agree. It’s that transitional environment of how do you help people first develop these new cloud models, how do help them optimize this traditional environment, and then provide the services that help this hybrid delivery model to evolve over time. And that’s a huge opportunity for HP & all of our enterprise services and technical services opportunities there are very significant indeed. But one of the things to think about long term is once you evolve to a full cloud model, like any utility, services don’t actually play a part of it. Think about how many services you use to run your electrical utility or your phone utility as an end-user consumer of that. It just works. And that’s a world which if you go out a couple of decades or more the ratio of services to the platform delivery of something-as-a-service is going to be radically different.
JF: People won’t have to have power stations attached to their enterprises any more.
DV: Do you actually think IT will get that simple in the next five years?
DS: I think it has to. I think that like any technology that evolves into a utility space that’s a natural evolution, because what you’re trying to do is offer your customer base, if you’re that utility, the best possible service level experience at the lowest possible transactional cost. Those are the two things obviously with a 100% available environment. There is no space in that for providing them consulting services to help them move along.
DV: Does that mean cloud’s the final resting place for IT as we know it?
DS: You know, that’s the way things evolve if you go to a true utility model.
Storage Vision
JF: The electrical power plant analogy is really relevant, because that’s fundamentally how they understand when they hear, “Can you imagine having a power plant attached to your home or business.” All that inefficiency. So speak to the inefficiency side of it. In the storage business you’re in storage obviously is where the action is right now. We’re seeing solid state going on, we’re seeing software being key. That stuff wasn’t around a few years ago. You guys were pioneering that at 3PAR. What’s you vision in terms of product leadership, where you guys will have that disruptive enablement within HP and for the customers?
DS: I think certainly in the area of the cloud there are a number of different problem sets; in fact a significant number of problem sets. One is providing storage that is really suitable as the basis for delivering enterprise IT-as-a-service. Clearly that’s where the 3PAR platform was optimized, with multi-tenancy capabilities, handling very unpredictable and diverse workloads, some of the performance scaling challenges, some of the needs for tremendous efficiency through the use of technologies like thin provisioning, some of the security challenges. HP has now the benefit of having really the premier platform in the entire industry to support that need.
But there are other challenges that are developing that are very cloud related as well and will increasingly become so. We talk today of the concept that “big data” wasn’t really descriptive enough of what is happening with this massive amount of unstructured data that’s being delivered, because it’s not just the data itself. You’re talking about the vast numbers of objects, the vast sizes of objects, not just the capacity, the vast amount of bandwidth to analyze the information and extract hidden information out of those objects. So again one of the things I believe HP has done really well is acquired modern core storage architectures not just with 3PAR but in the form of IBRIX scale-out file system technology to provide us with the basic assets that will allow us to solve customer problems in this new world.
Best-of-Breed vs Integrated Stack
DV: So I wonder if we could follow up with that and discuss one of my favorite topics. You and I talked a lot about best-of-breed versus integrated stacks. When you were CEO at 3PAR you made a very compelling case that “Hey, we’re 3PAR, this is what we do, we compete at our individual layer.” Then all of a sudden all these companies just went poof – disappeared – Isolon, Data Domain, 3PAR, Compellent. And it was really the case of a lot of large companies filling in their portfolios. So my question is can a company like HP or any large firm be both best-of-breed and integrated stack. What’s your angle on all that?
DS: Dave, I’ve argued for so long that as we evolve into this cloud or IT-as-a-service world those people trying to provide those cloud services will build their infrastructures off of best-of-breed infrastructure. And they have to, because they will be competing on the basis of what service level they can deliver at the lowest transactional cost. And if they choose any suboptimized elements in their infrastructure, they’re going to be at a disadvantage to another cloud service provider who chooses best-of-breed at every level of the stack.
Now within HP one of the reasons why I felt so excited about joining HP and continuing my career here is that HP has busily gone about selecting and acquiring the assets that allow it to rightfully claim best-of-breed status in all of the significant areas that make up the new cloud stack. So if you look at what happened in the area of networking you’re starting to see the dynamics of the networking industry change very substantially after the acquisition of 3COM, because you actually have a best-of-breed product now in the hands of a very large competitor to someone like Cisco, i.e., HP, where customers have that feeling that they want to buy from a big entity, but they want to buy a best-of-breed product. And 3COM’s technology as well as the ??? technology in each of their areas really delivers that best-of-breed.
The same has been true for a long time in the server world with all of the technologies that we have been able to deliver in blade systems, some of the exciting mixture of server and networking technology with Flex Fabric Virtual Connect. And what HP did with the acquisition of 3PAR as well as LeftHand and IBRIX is pull together best-of-breed assets in the storage world. So now you have best-of-breed in storage, networking, and servers, and you tie it together with best-of-breed orchestration and provisioning software with Cloud Systems Matrix. So literally you have this situation where the service providers delivering IT-as-a-Service literally can look at HP and at least at this point in time can say, “I can go for this entire integrated stack, knowing that I have best-of-breed elements at every point in the stack.”
Now obviously that may not last, and one of the elements of HP’s Converged Infrastructure Strategy is saying, “We’re open. We understand that if at some point in time we fall behind in some particular area you’ll want to substitute some technology.” And that’s very different from some of the other vendors of vertically integrated stacks who literally have packaged up their solutions & it has got to be exactly their product, even if that means choosing suboptimized storage or suboptimized servers in order to get that.
JF: Or storage you don’t want’a buy. You’ve got to get what they give you based on the stack, but you mentioned that HP’s acquisitions have been strategic, in this kind of LEGO block way in the architecture. You were a great acquisition. And they’ve done a good job. Have they had too much cuts in the past where it’s really cutting into the muscle? Is there internal invention going on in there from you perspective?
Converged Innovation
DS: Yes, I think there’s a tremendous amount of internal invention. If I just look in the storage space, all the StoreOnce dedup storage technology that we’ve brought to market was developed out of HP Labs collaboration with the HP storage division, and that is very exciting technology that we see not only exhibiting great growth rates but offering a very promising future. So a lot of internal development happening there. But if you look around the entire area, whether it’s the Virtual Connect connection fabric capability, the innovations we heard on Ecopods today – near-perfect PUE ratings – a huge amount of innovation.
JF: So they’re pumping the innovations in there. So you were an acquisition. So you’re saying you got an injection of more IT into the 3PAR equation, based upon you fitting into that puzzle.
DS: I absolutely believe so. Today we were able to announce the new HP storage strategy called Converged Storage. And it’s all about integrating in its two forms: Converged Virtual Storage targeted toward small-to-medium sized businesses and branch offices and then Converged Virtual Utility Storage targeting public and private clouds and large enterprises. It’s all about integrating these kinds of leading assets we’ve brought together, whether it’s 3PAR and IBRIX on the one hand or Left Hand, StoreOnce and IBRIX technologies on the other.
DV: …You’ve got the new OS, which is nice. It’s not a one-size-fits-all, but it is a one-size-fits-a-lot, and you’ve got LeftHand, you’ve got IBRIX, you’ve got StoreOnce integrated in there, and you’ve got some flexibility and choice for the customer. And then you’ve got a 3PAR and IBRIX integration….
DS: Which we also plan to integrate StoreOnce into over time,.
DV: It seems to me that that feels tactical in that it gets you to the market fast, but it seems like you’ve got a lot of legs there as well.
DS: Absolutely And what it does is it reflects the two different markets, because if you look at the cloud you really do need these multi-tenant specializations to handle very diverse and unpredictable workloads, security, etc., and that’s what the 3PAR element of the stack added to industry-standard design with its controller allows to deliver. On the other end for SMBs, they really want simplicity. If you’re in an SMB you really don’t want multiple elements of physical infrastructure that you then will have to network together. It will be very expensive, you’ll have to manage all of these elements. The same’s true for enterprises out in their remote offices or branch offices. So with Converged Virtual Storage what we do is leverage all of our blade system and Proliant DNA together with the scale-out software, the Store 360 Software, and the converged management with Cloud Systems Matrix, to allow people to run and co-locate for their applications as well as their storage software and capacity on the same physical infrastructure.
And that’s really radically simpler than our competitors who are offering people unified storage but then you’ve got to buy a separate compute platform, and then you’ve got to network it all together, out in hundreds of remote branch offices or in individual SMBs.
Unified Storage and 3PAR
DV: We used to get a lot of questions on unified storage when you were the CEO of 3PAR and a lot of Wall Street guys were looking at NetApp’s growth and saying, “Maybe you should copy that.” You always flat out resisted that and said your market was different.
DS: We resisted it for two different reasons. We were going after the public and private cloud space in its entirety, and we knew that you needed multi-tenant specialization for that space. We weren’t going after SMBs, which you’ll remember many of the discussions were about that we have had. So simplicity and co-location of applications and storage was not something that we had to deal with. Now, obviously, as part of HP with a much broader go-to-market segmentation, we have to provide excellent solutions, and we think we have all of the right applications to do that.
DV: So can a NetApp – to your point about a converged strategy rather than a unified strategy, a NetApp will say, “We’re going to go to market with a Cisco and a VMware, and we’re going to offer that capability.” What do you say to that?
DS: Well they can. I mean throughout history there have been lots of loose coalitions that have come together. working with coalitions to make things deliverable in a really integrated way not just once but also in every upgrade and update cycle is a very difficult challenge to try to meet. It really only works well when it’s all under one roof with one company.
The Place for Vertica
JF: So my question, Dave, going back to some of the inside baseball of storage, is the big buzz we’ve been covering is the hype around big data. You’ve joined HP with the acquisition, and Vertica is probably in the works. Donatelli said it isn’t part of his thing. Then I spoke with Andy Johnson at HP HQ a couple of months ago and said, “Why isn’t Vertica in the Storage Group? I mean Hadoop, storage, big data.” And he said, “We’re going to sprinkle Vertica around everywhere.” So is that the plan? Is Vertica more of a sprinkling strategy or is it more storage? Where is the right place for Vertica?
DS: I think the right place for Vertica is exactly where it is today. But I do believe there will be a sprinkling of those technologies, just in the way that we have all of the rich assets in storage that we have been able to bring together. We’ve got rich assets across the whole of the ESSN portfolio, networking, etc. Vertica provides a similar technology. Now again, moving forward, we have been talking about IBRIX’s suitability with its massive 16 Tbyte name space to be a huge, massive content pool for all of the unstructured data that’s being created. The advantage over all of these silos is that you don’t get into the complexity of managing all of these individual files, you don’t have the complexity of managing all of the data individually.
But we think that not only does it give you all of this tremendous scalability but it allows you to optimize cost performance points for data as it ages. And the problem with unstructured data will be exactly the same problem that hit structured data 5-to-10 years ago when everybody had 100% of their data storage on high-end monolithic arrays and they suddenly realized that 80% of it really should be on a much cheaper, tier 2 or tier 1.5 platform. The same’s now happening with all of this unstructured data that they put on expensive NetApp filers. And they are saying, ‘I want to put 80% of it on a more cost-effective platform that is more easily accessible. And that is what IBRIX will allow us to kind of achieve.
We talked today about the work we are doing with HP Labs developing a unique real-time content mining engine and being able to put that into IBRIX in the future. So we are going to have a number of different assets that will allow us to play in this big everything data state.
DV: Did you announce that?
DS: Yes. We talked about it in the press announcement today. But more importantly, one size is not going to fit all. Just as you have SAN and NAS solutions for storage, you’re going to have Vertica come on the analytics facilities as well as the metadata search facilities, you’re going to have Hadoop, ???, you’re going to have a lot of things.
You’ve got to be open, you’ve got to be a multi-vendor, you’ve got to have a heterogeneous set of solutions.
Hadoop and Dirt Cheap Storage
DV: And there is another dynamic I wondered if you could comment on in your roll here with your commentary on multiple solutions, extracting your knowledge.
DS: I thought this was a 15 minute interview.
DV: Is that what they told you? I wanted to get you to comment on Hadoop. When we talk to the customers who are running Hadoop the storage they are running on is dirt cheap storage. They’re getting a lot done. And that’s potentially very disruptive to the storage industry. So it’s almost the opposite of what you were saying about monolithic storage and the NAS storage. It’s the dirt cheap storage. Do you see that class of storage needing help, needing to be enterprise-ied, or do you see it as having to be dirt cheap, and what does that do to your business?
DS: I guess I didn’t make it clear. That was exactly my point in IBRIX. My point is you need a massive name space, something like a 16 Pbyte name space where you’re able to put dirt cheap storage behind and under this name space in order for a lot of this extraction of hidden value to take place. Absolutely looking at the different characteristics that you’re going to need. Some of this data may still have hidden value in it, but not all of the time. You have to store it for a long time, and you have to find as cheap a platform as you can to store it. And you have to be able to move it back to more high performance platforms if there’s a demand for it in the future. You never know, especially in fields like medical research, when some old data may suddenly be the key to a new discovery.
DV: And the economics are there? You’re on record as saying in the Cube that you’ll be able to compete from an economic standpoint with that trend and with that IBRIX vision that you put forth.
DS: What I said is that we certainly understand what the trends are, and we’re certainly looking top make sure we can play a major part in the evolution of the industry.
JF: Thanks for coming back to the Cube. Great to see you again.