SEPTEMBER 11, 2003
SAN FRANCISCO, CALIFORIA
© Copyright 2003 Hewlett-Packard Development Company, L.P
All rights reserved. Do not use without written permission from HP.
Good morning. It's a pleasure to be with all of you today.
We meet this morning on an anniversary that none of us will ever forget. Exactly two years ago, at this very moment, all of us were riveted to our televisions and radios, watching a series of truly horrifying events unfold before our eyes in New York, and in Washington, and in Pennsylvania. And as we already said then, one of the terrible truths as all of us watched is that some of the things we hold most dear in this country were used against us. Our love of freedom and freedom of movement was used against us; open borders were used against us; and perhaps worse for this industry: our best technology - from the Internet to airplanes - was used against us.
And of course, in the days and the months that followed, it became increasingly clear that our world today is caught in a race between the organized forces of construction and the organized forces of destruction; between those who want to build things for our fellow citizens, and those who exist to tear things down.
We talk a lot this week about important things like open standards and business agility and return on investment and innovation and I will talk about all those things today as well. But I think it is important today that we stop and remember that we - particularly, we in this industry - should never forget that what we all do doesn't just help make a dollar for our companies, and a dollar for our customers; what we all do make a difference for this country and for many others.
I think at the root of this important work we all do is partnership, and this event, actually, is all about partnership. Beyond the 80,000 joint customers that we and Oracle share across every continent, our partnership with Oracle even extends to the high seas. Some of you may remember that at last year's address, I mentioned that it was HP that helped put the wind in the Oracle-BMW Racing Team's sails. And we were delighted that Larry was a finalist last year for America's Cup. Next week, Oracle-BMW goes up against Alinghi for a rematch of America's Cup right here in the Bay Area. Last year, I asked where Larry was; this year we all know where Larry is, and Larry is on the home seas, so the pressure is really on.
Based on the commentary and the meetings from this context, I think the good news is that the industry is beginning to converge on a common goal: that is to create an open standards-based technology infrastructure that will truly deliver on the promise of better reliability, better availability, better business agility, better security, and a better return on your information technology investment. I also know that you have heard from a number of my esteemed colleagues this week on the subject of next generation data center technology and grid computing. Obviously, that's very important, because we believe that grid is actually the next major architectural transformation of IT. Now, from the sound of some of what was said this week, it appears that some people may be trying to ride this horse before they are ready and as we found out earlier this week, that is a good way to hurt your foot.
One of the big problems with grid - just like every other technology inflection point - is grid actually has been more hype than reality, so far. There's been more hype about what grid actually is, and what it will really take to make it real. If you do a Goggle on grid, you will come up with more than six and a half million hits, so when I hear other people preach about the path to grid and say that standardization is the key, my hype meter goes off.
I actually can understand why you would talk about standardization if that is all you can deliver. But, we all know that the next generation data center is about standardization for sure, but we also know it is about a whole lot more. What about virtualization and utilization? Or, as Larry says, the technology required to create the illusion that this is just one system? Because whether it is across multiple data centers, or within a single data center, it is about technology transparency. What about the management software that's required to gain visibility and control over the entire system, and all of the component parts, including the processes and the applications that run on top of it?
What about the innovation that's required to automate much of the processes involved in operating a traditional data center - right down to monitoring and measuring the heating and cooling requirements? One little point on the subject to standardization: it is also true that some of us here this week think it is important to both scale out and scale up with standards-based offerings.
My hype meter also goes off when I hear somebody claim about anything in the enterprise space - but particularly grid - that R&D is overrated. It brings to mind what our friend Craig Barrett at Intel said in response to that recent Harvard Business Review that claimed IT doesn't matter. What he said was that 'the only people I hear who suggest that Moore's Law doesn't matter - the people who say R&D doesn't matter - are those who don't want to keep up or can't keep up.'
My hype meter also goes off when I hear somebody claim that companies should buy all of their data center from one vendor. True grid is not a homogeneous, one-vendor-can-supply-it-all proposition. It is a community effort, and like the open source movement, the community benefits from more - not less - collaboration and diversity.
If you want to have a real discussion around advanced technology and next generation data centers, the lesson is you'll have to talk to a real technology company.
Finally, when asked how their company embraces grid today, my hype meter goes off when I hear somebody stand on a stage and actually reply that they've implemented a RAC with a lot of servers. Now, we are a company that actually operates a commercial utility data center. We run our R&D environment on one. It is the equivalent of grid functionality inside an enterprise. It is a highly tuned, resource sharing, reliable, secure, automated data center of the future - running today.
Let's have a real conversation about what it will take to deliver grid technology across commercial enterprise companies…
To start with, we are taking about at least a three- to five-year journey to create a commercial enterprise-class grid; one that actually operates across multiple heterogeneous data centers. The grid will require technical innovation from a community of companies and developers in the order of thousands and that's even tens of thousands in terms of engineers who will work hard to solve the hard technical problems.
And those hard technical problems can only be solved with real enterprise systems R&D, real engineers, real technology, real partnerships, and real service processionals. Grid is about a whole lot more than a single RAC and a lot of servers. The grid requires a true commitment to engineering and R&D to commercialize the promise of grid. The reliability that many of you have come to depend on running Oracle solutions on HP platforms - which has made us Oracle's largest multi-OS and platform partner, bar none; that reliability now has to extend to grid.
Last week, to compliment Oracle's 10g launch, HP announced our plans to grid-enable our entire product line to help move grid computing beyond research, to true commercial applications. Grid architecture, however you define it, will be embraced by companies in their quest to create an environment where business priorities drive IT, not, the other way around…where IT is perfectly synchronized with the business.
Infrastructure has to be able to flex with the business, whatever the business demands of it. And so, IT has to become a platform for the efficient and effective operation of the business, a platform for the efficient and effective delivery of data and business applications and business processes. So grid becomes a really important technology to achieve what we call the Adaptive Enterprise.
In May, we introduced something called the Darwin Reference Architecture, which shows how HP brings together Web services, grid services, advanced management capabilities, and virtualized resources to create the Adaptive Enterprise. We named the reference architecture after Charles Darwin, because it is an evolutionary path for customers to build on open industry standards, on a modular set of building blocks, and these blocks are designed to be interoperable and heterogeneous, and to work with legacy environments.
The architecture is open to all partners, and it is open to competitors - which may be good news for many of you, particularly if you are working with competitors who can't show you an architectural blueprint with roadmaps, shipping products, utility data center implementations, and real customer success stories. The grid requires grid-enabled products and technologies.
Today, HP already grid-enables our PA RISC, and our Integrity, Alpha, and ProLiant server lines. We do this by incorporating the Globus Toolkit and Open Grid Services Architecture 2.0. And over the next 12 to 18 months, HP will integrate the Globus Toolkit as well as the next version of OSGA Standard 3.0, which is due out in October, to our entire commercial line.
To bring grid to the enterprise, we will also leverage grid standards, helping customers simplify how they use and manage distributed IT resources. And, because we happen to be the largest consumer technology company in the world, we are also grid-enabling our consumer products and systems all the way from the smallest handheld devices to PCs and printers.
Delivering grid also requires a services organization that can design, and implement, and operate grid solutions. HP Services - which is 65,000 professionals strong, worldwide - now has a dedicated group that specializes in enterprise group consulting to help customers get grids up and running. Among other things, our consultants provide management, deployment, and life cycle support for grid environments - again, all with the goal of extending our Adaptive Enterprise strategy to make IT systems more responsive to customer needs.
But we actually do agree with one thing that we heard in Monday's keynote: the way to reach the nirvana of computing is through systems management. But we also believe that delivering grid is about more than simple device and application management. It is about managing the hardware infrastructure, business applications, and networks. We are building on the HP OpenView platform. HP OpenView is already used by every single Fortune 50 company but one - you can imagine which company that is. And we are extending the capabilities of that software up through the management of Web services, so that we can deliver comprehensive, real time business process intelligence, and enable immediate IT resource response in the context of Web services or grid services.
The goal of the grid is to be able to treat all IT resources as services. OpenView, through our leadership with the Web services management framework becomes the enterprise management platform that can manage all of your enterprise services and resources. And, through our Utility Data Center, which already uses the Globus Toolkit as the backbone to link all the hardware and software in a network, we are going to extend our experience as the grid develops.
We have been privileged to work in collaboration with some of the best research institutes in the world on grid, and let me just name a couple… Our partnership with the U.S. Department of Energy, for example. This is where an HP supercomputer will soon be the largest computer attached to the grid; or our partnership with CERN'S Open Lab for data grid applications; or our partnership with Biomedical Informatics Research Network, where HP systems are in use today for grid-based life sciences research, or the Pittsburgh Supercomputing Center's Teragrid, which will provide the nation's fastest and most powerful computing grid.
And there are, as well, other key partnerships to advance grid in the commercial market. Together, HP and Oracle are working with leading consortia, including OASIS, W3C, Globus Labs, and the Web Services Interoperability Organization to drive industry adoption of open standards for grid computing. As the leaders supporting Oracle 9i RAC with the price/performance benchmarks to prove it, we think it is a natural evolution for us to become the natural leader for 10g adoption. HP's grid strategy supports AE by delivering IT, including Oracle's 10g, as a service transparent to the user, no matter where that service is coming from.
And ultimately, the grid ought to be able to render almost anything in IT - computers, processing power, data, Web services, storage space, software applications, data files or devices - as grid services. In the future, where this will hopefully evolve is for enterprise grids to be linked together to create grid computing resources across data centers, even potentially linking grids around the globe. If grids ended up as a set of proprietary, non-interoperable islands it is counter to the fundamental promise of the grid-envisioned, loosely coupled, service-based IT world.
Imagine what would happen to the Internet if it had been built on proprietary, non-interoperable technology. HP's goal is for the grid to do for IT resources what the Web did for documents - provide ubiquitous and easy access.
Now, of course, we are many years away from that vision. From where we sit, there are five basic challenges today that need to be solved on the way to achieving the ultimate vision of the true grid with IT delivered seamlessly as a service. The first challenge is the complexity of grid software. As we all know, existing Globus Version 2 is almost exclusively about, and for, scientific and technical computing. Globus Version 3 is for commercial and technical computing, where grid meets Web services, and the grid becomes much more powerful, and I think that version is due to be released in a month or two.
But, Globus Version 3 is a very big block of code - and growing. The sheer size and complexity of it increases the risks associated with failures, and with errors, and with deadlocks, and security. And those risks need to be worked out before a CIO can get close to considering it. This has to be easy to use and manage.
Challenge number two is trust and security. The issue here is not the security of the bits on the Internet, it is about the security required when sharing applications across data centers. Certifiable trust is a base requirement before a CIO will consider it. Challenge number three is heterogeneous operation. Given the multiple number of operating systems, and platforms, and data formats, and applications, interoperability is a given, but it is also a major challenge.
Challenge number four is fully open standards. As we have been saying all along, open standards really are essential to realize the vision of true grid. And challenge number five is robustness. Globus Version 3 is bound to have bugs, and those bugs of course, have to be addressed. Fortunately, there is a very large body of developers in the open source community working on this problem, but we have to come up with ways to identify problems and automate the fixes.
So, what's our point of view on how to address these five hard and real technical challenges? We are focused squarely on the management and execution of grid services. Let me say that again. At HP, we are focused squarely on the management and execution of grid services. This is where we intend to make a major contribution. In June of this year at the Global Grid Forum 8, we demonstrated three HP-invented technologies as part of our contribution to the grid community.
The first was something we call the Grid Topology Designer, which is a simple graphical interface for pulling together multiple data centers into an enterprise-class grid. The interface leverages the millions of lines of code that we have invested in our own Utility Data Center product for delivering IT as a service within the enterprise, and turns it into a powerful resource for brining multiple data centers together. In other words, it is kind of a grid on steroids, and it reduces complexity, heterogeneous support, and it adds robustness and security to the grid. And it does so with an open standard interface to grid built on grid protocols.
Secondly, we've contributed what we've called Smart FROG technology, which stands for Smart Framework for Object Groups, to the GGF and the Open Source community. Smart FROG, allows customers to program the grid and reduce complexity by minimizing configuration and deployment requirements.
And number three, we contributed a Web Services Management Framework. The framework is open, it's standards based, and we have contributed it to OASIS, and it helps us manage the Adaptive Enterprise platform for our customers. We are grid-enabling it to make it a Grid Services Management Framework, and again, this is about reducing complexity and supporting heterogeneity and open standards.
We think these contributions all demonstrate progress towards enabling an enterprise CIO to one day actually manage and execute grid services. Oracle 10g is making a significant contribution towards bringing grid services into reality. We all know that clustering technology has been around for about 20 years, but this is the first time that anyone has come up with a commercial application within an enterprise at the application level, and Oracle deserves a great deal of credit for this, because application-level grids in the Adaptive Enterprise framework are a significant contribution.
So, what do HP and Oracle offer on Oracle 10g that is different from others? We actually think it's pretty simple. The HP platform is key to Oracle's grid initiative. As we all know, HP already offers the broadest selection for Oracle environment with a 41 percent share of the total market for all operating systems and that share is growing; that includes market leading positions in UNIX, in Linux and Window servers tuned for Oracle environments, storage systems tested in Oracle, and management software enhanced for Oracle environments, and appliances that enable enterprises to publish data or access data from remote locations.
HP is the multiplatform infrastructure, blades, and price/performance leader for Oracle environments. In fact, HP has more Oracle 9i RAC installations than any other vendor in the world, and more customers are running Oracle 9i RAC on HP platforms.
Earlier in the week, you heard one of our competitors talk about their work with FedEx as it relates to scalability. Most of you know, I hope, that FedEx is one of HP's marquee customers and partners. FedEx's application servers - the systems that FedEx relies on to deliver their transformation and information solutions, the systems that require massive scalability, the systems, in other words, that they actually rely on - these run on HP servers. And it doesn't stop there. FedEx recently awarded their most strategic package tracking project, called Fusion, to HP. And while I can't reveal the proprietary details of this project, suffice it to say that it is running Oracle 9i RAC on HP UX in the back end, and Linux on HP ProLiant servers on the front end.
I am also proud of the work that we are doing together to expand our Itanium initiative, and the work we're doing together to build a worldwide Oracle application server middleware agreement that will deliver more 10g solutions to our joint customers.
We have been doing joint R&D and the co-development of technologies and services with Oracle over this 20 year period. For example, we worked together for five years to create the Oracle 9i RAC. HP and Oracle strive to continually deliver industry-leading price and performance. So, we actually think there is no other platform partner that has the technical breadth and depth in terms of our work together.
Just this week, HP and Oracle announced new industry leading benchmarks for 10g. A year ago, HP and Oracle partnered to achieve the world's first Linux on Itanium benchmark, and today, we follow that achievement by announcing the world's first Linux on Itanium TPC-C benchmark, using a four-way industry standard integrity server with Oracle applications for 10g. This makes HP and Oracle a powerful combination that outperforms all four-way servers in its class, and offers better price performance than IBM's 8-way eServer. We also achieved a new SPECJAppServer2002 benchmark using Oracle 10g on ProLiant servers running Linux.
Now, beyond great performance benchmarks, what makes me believe that the combination of HP's Adaptive Enterprise strategy and Oracle's 10g strategy is going to continue to be the best combination going forward?
Because we will continue to lead the way with a commitment to creating the open standards- based infrastructure that you have heard so much about over these last few days; because we will continue to lead the way to bring modular architectures, like blades and storage, into the data center, while some of our competitors ignore some of these new technologies; because we will continue to lead the way in both .NET and J2EE applications and services; because we'll also continue to lead the way in engineering systems and infrastructure that accommodate and embrace heterogeneity in the data center.
It's because we will continue to lead the way with offerings engineered to deliver a better return on IT, not just in the data center, but also across your PC, your handheld, your wireless, and your printing infrastructures; because we'll continue to lead the way when it comes to rethinking how information and data flows across your enterprise in all of its forms - whether it is paper or digital - and will continue thinking and working on how to mesh those systems together in one seamless adaptive infrastructure.
And because we'll continue to lead the way in moving to a single Itanium chip architecture across the entire server family, because supporting a multi-chip strategy - whether you are a company managing an IT environment or a vendor like IBM - it is a costly value proposition. Because we'll continue to lead the way when it comes to driving the adoption of industry standards through the more than 700 HP people who participate in 300 different standards organizations around the world. And because we'll continue to lead the way in helping key Oracle 9i database partners and customers migrate to the next generation Itanium processor family. And that's true across Windows, .NET, Linux, and HP-UX operating environments.
And as I mentioned earlier, we are already seeing significant performance, CPU, memory, IO, bandwidth, and graphics capability improvements on the Itanium benchmarks that we're running. Because we'll continue to lead the way when it comes to delivering the industries' best performance benchmarks for Oracle databases, applications, and middleware running on Linux. Today, Oracle 9i is delivered across our Windows, our UNIX, and our Linux platforms - soon to be followed by 10g. And this, I think, is an important example of our fundamental bet on heterogeneity and our bet that being the best platform provider across a heterogeneous environment is what our customers are looking for.
We're also going to lead the way in automating the management and control of your entire IT environment - all the way from the underlying infrastructure to the applications, and the business processes that run across it. And we think about the environment in the way I just described - horizontally, with business processes, and applications, and IT infrastructure.
In the area of improving manageability, we are betting that next generation infrastructure management will become one of the most important criteria that you will use when you select IT systems and partners. And we're the leader in manageability with OpenView and we intend to continue to expand on that lead, both with organic investment and with acquisitions, some of which we have already made.
When it comes to Oracle, HP OpenView is a flexible, solution-ready architecture that combines readily with smart plug-ins, third party software, internal software, and industry standard integration buses, and looks at the entire network for the Oracle stack.
And HP OpenView also provides the best and most complete, end-to-end, multi-platform management solutions which lets companies manage centrally from a single pane of glass by using a smart plug-in for Oracle. This week, Oracle and HP announced the simplified and effective management of resources across the entire IT infrastructure, with the HP OpenView smart plug-in for Oracle, which helps administrators monitor distributed, enterprise-wide Oracle environments from a central, best-in-class console.
And in the area of better utilization, we are investing in a range of virtualization products at the server, storage, and data center level. In the server environment, we have the leading virtualization capability from the low end to the high end; from ProLiant industry standard servers to our OpenView software, to HP-UX's award-winning partitioning continuum. In a storage environment, we have everything from the appliance-based approach, to storage virtualization that customers can use to get started very quickly - all the way up to ultra-scalable, feature-rich environments and implementations. And we have achieved network- and data center-level virtualization through our Utility Data Center offering which you just saw.
UDC is the data center level virtualization set of capabilities. By the way, this is not us talking, this is others talking. So Gartner, for example, has estimated that we are 18 months ahead - at least - of both IBM's On Demand and Sun's N-1. Now, once again, HP is the only multi-platform infrastructure, blades, and price performance leader for Oracle environments. HP disk arrays use Oracle's SAME - Strike and Mirror Everything - technology to achieve higher performance and better efficiencies.
Now, when all is said and done, what it all means is that HP and Oracle are better together for our customers and for the future of this industry. After two decades of collaboration, it is not an accident that we have both come up with a strategy that is practically joined at the hip. We really do believe that grid is a big deal. It is driving the R&D and innovation engines in our industry. And for the first time, our energy is focused on inventing something far more radical and far more profound than a killer app, or a hot box. Now, our energies are focused on making computing more useful, more compatible, and less expensive to manage.
Business decisions ripple instantly and easily across your organization. IT and business perfectly synchronized. Imagine what's possible when that happens…and by the way, just to close, who said R&D is overrated?
Thanks very much.