Senior Vice President, HP
Director, HP Labs
Address at Software 2005
Sand Hill Conference
Santa Clara, California
April 27, 2005
© Copyright 2005 Hewlett-Packard Development Company, L.P.
All rights reserved. Do not use without written permission from HP.
It’s a pleasure for me to be here this afternoon.
I think there is a perception of researchers that we spend
our days locked away in a basement somewhere. But they do
let us out from time to time. For that, I want to thank you
for the opportunity to be here today.
About six hours ago, our chief strategy and technology officer,
Shane Robison, started the day with a provocative statement:
that while HP is well known for our hardware -- and more
recently, as one of the largest global providers of IT services
-- we are not as well known as a software company. Part of
that has to do with the fact that we did not have a focused
software strategy. As it turns out -- in Nora Denzel’s
memorable phrase -- practicing “random acts of software” is
not a good way to go.
We've changed quite a bit. As you can in see in both our
internal organic investments as well as in our merger and
acquisition activity, we've been quite aggressive recently.
Now, after hearing from HP's Shane Robison, Russ Daniels
( Chief Technology Officer for our software business), and
Martin Fink (who runs our Open Software business), I hope
you have a clearer idea of HP’s software strategy and
the level of innovation we are bringing to our customers
and our markets today.
Software business growing for HP
Software is not only a substantial business for us, but
a growing business. HP today has a $3 billion enterprise
software business. Our OpenView network and system management
software is used by every single one of the Fortune 50 to
manage increasingly complex IT networks.
Our OpenCall software is used by more than 200 telecommunications
companies – including 40 of the top 50 worldwide – to
bring reliable and secure services to hundreds of millions
of users. Eighty percent of the world's SMS messages flow
through HP systems.
And an increasing number of enterprises today rely on our
mission-critical operating systems, our clustering and high
availability software; and our broad portfolio of storage,
networking and communications software.
We hope that you’ve taken away that we have a unique
strategy, focused on open, distributed software. Like many
people, we've focused on a future built on software as a
service, with grid and utility service models for delivering
that. At the same time, we are not focused on building vertical
stacks of proprietary middleware applications and databases.
We also hope that we’ve made clear that our approach
to best-of-breed software is to focus our innovation in software
on areas where we can add the most value, where we can make
a difference, and then to work closely with industry partners
to fill the rest of the pieces we need for our customers.
That’s our business strategy, and that strategy also
shapes our work in research.
Largest element for HP Labs
As software becomes an increasingly large part of HP’s
business, it has also become the largest element in the business
of HP Labs.
That said, I’d like to tell you about some of our
software projects in HP Labs. As a company, our big ideas
in software are focused on the infrastructure side. We are
not trying to build the next iTunes or other software that
is consumer-focused. The question we’re trying to answer
today is: what is the software infrastructure required to
help businesses today, and what will be needed as these businesses
move into the future?
These projects range from ones strongly aligned with our
business units to ones out on the frontier. Much of this
new technology is being shaped differently than we've done
in the past.
In many cases, we are in strong partnerships with cutting-edge
enterprise customers around the world because we're seeing
increasingly what drives a lot of software innovation is
understanding where there's an intersection of the technology
capabilities we have with critical business problems that
enterprises are facing..
Inventing for HP
Before we get into the projects, let me give you some background
on HP Labs.
As one of our Senior HP fellows, Alan Kay, has said, “The
best way to predict the future is to invent it . . . ” That’s
why we exist. HP Labs was founded in 1966 when Bill Hewlett
and Dave Packard decided that HP needed to create a group
dedicated to a longer-term view of technology issues while
the business divisions focused on the day-to-day battles
they had to fight. That was true then, and it's even more
On one hand, we work with our business groups to develop
technologies that create competitive advantage for them --
technologies that will help drive their road maps for new
products and services. In some cases, we jointly define the
strategies and road maps with those businesses. At the same
time, our mission is to deliver breakthrough technologies,
ones that define that whole new opportunities for HP.
In other words, while we have some encampments on the fuzzy
frontier of technology’s future, we rarely work on
technology for technology’s sake. What we focus on
is where these technologies will make a difference, what
customer will value it and how HP can build a business around
Shift in focus
If you look at HP Labs programs over the years, you will
see a fundamental change in how we’ve chosen to invest.
For the first 25 years of our existence, most of the innovations
coming out of HP Labs were primarily hardware and device
innovations, because there were many opportunities based
on unique hardware technologies.
These included things like the world’s first pocket
scientific calculator, light-emitting diodes, thermal inkjet
printing and HP's precision architecture RISC technology.
All came out of HP Labs. There are still many opportunities
based on hardware and device innovations. But more and more
we are shifting our attention, looking at high we can build
high-value software that allows us to leverage industry-standard
That's evident if you look at some of our innovations of
the past decade -- our 64-bit architecture, something we
call the digital media platform (which I'm going to talk
about in a moment), photographic-quality printing and low-cost,
better-than-film-quality digital cameras, model-based automation
tools, shared applications servers, automated storage management,
the first Utility Data Center and utility computing services
such as our recently deployed Utility Rendering Service for
the media industry.
Meeting business needs
Why the change? Put simply, it’s what our customers
value. Even in the midst of the Internet bubble, CIOs were
telling our research team that they were under increasing
pressure to control costs and deliver more business value
from IT, to be more agile and responsive within their companies.
Around 1998, we started to shift our focus. We moved out
of some areas, like microprocessors, that were not giving
us sufficient differentiation. And as we began to sort out
our software strategy as a company, we increased our investment
in software for the Adaptive Enterprise where we knew we
could have more impact.
Today, in HP Labs, we are focusing our innovation in six
fundamental areas. First, the area I just described, what
I call reinventing the economics of IT. A grand title, but
that is, in fact, our goal. This includes the Adaptive Enterprise,
our Grid and utility computing work.
Second is printing and imaging growth in new areas. As many
of you know, HP is fabulously successful in that area, but
our real goal is to expand the footprint of HP's printing
and imaging business beyond the desktop at home and in the
Third, digital entertainment --not just the devices you
and I see in our homes, but also the infrastructure, the
new backbone being built for that industry both in distribution
and content production.
Fourth, technologies for services delivery. Most people
think of services as a business that delivers technology.
We do that, but we're also exploring how we can change the
services business through the application of technologies
-- the same way IT has changed manufacturing from being about
capital and labor to being about information and inventory
Fifth, vertical industry solutions. These are high-value
technologies we can develop for industry segments such as
communications, media, manufacturing, financial services
and so on.
.And finally the sixth strategy is in what we call disruptive
and emerging technologies that are not part of the current
What all six of those strategies have in common is that
software is at the heart of every one of them. I'd like to
tell you about three specific programs where we are contributing
to HP’s software strategy today.
Next-generation management software
Our first and biggest area of focus at HP Labs is next-generation
management software. You heard Shane and Russ talk about
the Adaptive Enterprise Grid and utility service models.
The Adaptive Enterprise is sometimes difficult to wrap your
arms around, because it is actually three things at the same
time: First, it is HP’s view of where information technology
is going. Second, it is a company-wide program that covers
services, consulting practices and products. And third, it's
about some unique technology contributions focused in this
One of these technology components revolves around determining
how we deliver a utility for computing. This is another area
where there's confusion because sometimes people talk about
utilities the way that they talk about grids, as this free
resource that we're all going to use. One model of utility
computing is in fact a provisioning capability, a public
capability to provision IT.
But at other levels we see it as specialist services to
provide dynamic outsourcing to specific industries or internal
to corporations where people are looking for ways to more
flexibly use the IT resources they have. And unfortunately
a lot of times people go for the more Utopian side of that
equation and don't realize that in a practical level when
people talk about a data canter consolidation actually a
lot of the elements of utility computing are happening today.
So in an Adaptive Enterprise, in this emerging service-oriented
architecture, management software has to do a lot more than
just collect and present data. You'd have to integrate information,
to automate processes that are manual today. It needs to
provide closed loop control. How can we build a system that
doesn't require operators any time anything changes in the
Making change easier
To us, the irony of management software today is that while
business change is rapid, changes in IT systems are anything
but. If you are introducing a new product, or making an acquisition,
or merging two divisions, IT is guaranteed to slow you down.
The question for us was: How can we make IT more responsive
to those changes? The answer is policy-driven, model-based
automation. Let me give you one example of how we’re
tackling this question.
In HP Labs, we’ve come up with a technology that was
recently announced as part of our OpenView Automation Manager.
Internally, we call it Quartermaster. It is model-based automation
tool whose goal is to lower labor costs by taking labor out
of the system, to lower capital costs by better asset utilization
and to help customers run more robust systems by reducing
the amount of manual operations that take place. And it gives
us the ability to automate the response of an IT environment
to changing workloads.
We know that some of the biggest problems in data centers
today are operator-induced. Given the large number of changes
that take place in any data center, even skilled operators
make mistakes. Just one extra comma put in the wrong place
can make a whole system fail. These errors are very costly – and
for most businesses, IT systems are an integral part of keeping
their business running.
Just imagine an IT manager getting an urgent request that
says: we need an e-commerce system that is capable of handling
peaks of 30,000 queries per hour for a sales promotion starting
two weeks from now. Today that means IT operations, IT architects
have to build a system. They're writing scripts, they're
modifying and editing things. All of these things can take
a long time to do, and more time to verify that they're right.
Automating IT change
With Quartermaster we have libraries of models for all the
key system elements. We have templates for key kinds of applications.
An IT designer can simply drag in an e-commerce site on a
palette available from a resource utility and enter requirements
Instead of a process requiring many weeks to build a new
system, the system's description can be created in less than
an hour. The deployment of it -- that is, the compiling of
that specification -- can be done in a few hours. From that
to a deploy test cycle can take less than a week. The result:
a dramatic change both in time required and the level of
confidence that people have when they deploy a system.
I mentioned earlier that we are a major telecom supplier
globally, and the Labs has worked on that. If you think about
a telephone switching center, you picture in your mind endless
rows of faceless machines and no people. In building software
for that environment, HP has developed a lot of tools that
bring extremely high level of automation -- libraries that
can be debugged and then reused over and over again, policy-driven
models and more.
That's what our telecom experience has given us and we've
applied a lot of that same learning to Quartermaster and
some of the other tools that we've developed. IT automation
is all about time saved, money saved, labor taken out of
the system and dramatic improvements in reliability.
Also that level of automation also makes things possible
for customers -- it allows people to experiment, to retune
and redesign websites and services that, if manually deployed,
would just be too risky and too time-consuming. So that's
one of the other benefits that we get.
Another area where HP Labs has focused its attention on
the Adaptive Enterprise is on the issue of security. We have
a very simple goal when it comes to security: to secure access
to the right information, by the right people, all the time.
Of course, traditional solutions to the security challenge
include analyses of networks and systems and the installation
of firewalls, intrusion detection systems and anti-virus
software. But as we can see given the challenges we face,
those tools are not giving us the answer we want.
At HP, we're looking at new approaches.
A program we call HP’s Active Countermeasures, for
example, is a distributed scanning tool that imitates a computer
worm’s ability to take advantage of vulnerabilities
on certain machines, and gets to those machines first. We
deployed this back in the period when Ninja and Code Red
were infecting a lot of corporate IT systems. We actually
deployed our own worm on our corporate network with the agreement
of our own IT people.
What that allowed us to do, in fact, was to minimize the
damage at HP in some incidents because we could identify
systems early on and either notify users or we could take
systems offline. So we are using that and in fact gained
a measure of advantage from it, and we are now taking this
technology to market.
Another program, “Virus Throttler,” actually
mimics the body’s ability to limit damage from unknown
A lot of the problems we have with security today stem from
the fact that we're in a race. Microsoft releases a patch,
virus writers analyze it before you install the patch. Now,
well-informed as to where the weakness is, they engineer
a virus to attack. It's a cycle.
What we at HP Labs did was explore what signs and symptoms
are signatures of attacks. And in fact there are a number
of system parameters you can measure to start identifying
the presence of a virus even if you know nothing about its
Viruses need to multiply quickly, so we see systems where
a virus may be trying to make 1,000 machine connections per
second – very much abnormal behavior for any machine
that we've studied. We respond by choking off that message
flow. We don't actually shut the system down, but we reduce
throughput to that system enough so that it gives IT staff
time to respond.
In a way, it’s a corporate immune system for computers.
And it is now shipped with all HP blade servers. It’s
also available in the HP ProCurve Ethernet switches.
Keeping systems safe
Of course many viruses run when you launch an email attachment,
open a file with macros, or visit web pages that use scripts.
Unlike some malware that depends on security holes in a piece
of code, these kinds of viruses aren’t exploiting flaws;
they’re using the system the way it was designed to
We have a program, with the internal name of Polaris that
is designed for Windows XP to stop virus attacks on standalone
devices. This innovation from HP Labs allows users to configure
most applications so that they launch with only the rights
they need to do the job the user wants done. This simple
step, called enforcing the Principle of Least Authority,
gives so much protection from viruses that there is no need
for the standard pop-up security dialog boxes.
The Alpha release of Polaris is available now under a controlled
roll-out. We have users at HP Labs and have started customer
Lastly, in information systems, one of the key security
issues is: how do you know that your data was never compromised?
Services delivered by utility computing require increasingly
complex infrastructure, automation, and administration processes.
But why should customers trust that their transactions will
be secure, or their confidential data sets will remain protected?
In HP Labs, we have developed a technology we call “Trust
Record,” that does a complete audit of your system
at all times, and generates a monthly report that looks a
bit like a utility bill, showing what happened, and what
actions were taken. It helps businesses comply with all the
regulatory compliance and other transaction security issues
of Sarbanes-Oxley, HIPAA and many others.
Software to manage networks
Another area we are building software assets in is managing
networks and computing used in digital media applications.
This is one of the keys to HP’s digital media strategy.
The entertainment business is changing as two businesses
collide. On one hand, you have traditional media and entertainment
companies. And on the other, you have network and service
providers. Both industries see enormous opportunities.
There are huge problems in providing a software infrastructure
to enable a lot of these new capabilities and do it in a
way that provides the security that content owners expect,
at the same time it delivers the economic efficiency that
content owners and network providers need.
Digital Media Platform
HP is developing software that builds protection and connections across
those networks today. At the core of this strategy is a technology we
call the Digital Media Platform.
One of the problems today is that too many media and entertainment
companies are stuck halfway between the analog and digital
worlds. For movies, once content is shot, even if it starts
out as digital media, it’s still a multi-step, multi-week
process to make sure it’s logged, tagged, and shipped
properly – which is time-consuming and expensive. Once
it gets to post-production, even if it is in digital form,
it has to be converted back to analog, just so key stakeholders
-- from directors to investors – can view it.
But rather than creating content many times, and then distributing
it many times – imagine if there was a way to create
content just once, and then distribute it wherever it needs
to go – whether it’s a movie theater, an airline,
pay per view or a film festival. Imagine if you could do
all the work necessary throughout your process with just
one window, a digital dashboard.
Automating digital workflow
That is what we announced a year ago – to build a “create
once, distribute many” platform for the media and entertainment
industry, and to build it on an open industry standard foundation that
would help all entertainment entities, from TV to movies to radio, make
the full digital transition a reality. The overall goal is to help automate
the digital workflow, making the entire end-to-end process faster, more
reliable, more flexible, more secure, and more cost-effective.
So, the HP Digital media platform is an industry-standards-based
framework of enterprise software, hardware and services which
allows media companies to digitize, store, process, manage,
distribute, and archive complex media assets securely and
efficiently. Just one week ago at the Conference of the National
Association of Broadcasters, we announced that with our partners
at Ascent Media Group, the first real-world implementation
of the HP DMP is up and running at Sony Pictures Entertainment.
As proof of how well it works: by mid-April, they had hoped
to digitize 100 titles using the DMP. Instead, they’ve
done over 600.
We are now working with our partners at Warner Brothers
to use the same technology to complete a digital end-to-end
process that will save time and money. You will soon see
other media companies added to that list.
Looking into the future
The innovation I’ve talked about so far – from
automation to digital media – is being developed and
deployed today. But we’re also looking at the future – sometimes
10 years or more out – asking what the world might
look like then and the software innovations that could redefine
the ways we do business.
One interesting area of research is information services.
The idea is to use software to break through all the clutter
and cognitive overload in our information-saturated world
and find those things you care about.
Another question we’re asking today is this: in this
connected, networked world of the future, how do we efficiently
allocate networked computing resources? If we really do move
to a grid or utility model, where computing power is delivered
as a service like electricity by a central provider, how
do we deliver it?
Allocating resources with Tycoon
As long as the number of users is small, the problem scales
well and resource allocation can be done socially. But as
soon as the grid becomes large, any such system is likely
to break down as too many people compete to use the same
resources. It’s easy to imagine that valuing and prioritizing
billions and billions of transactions will be impossible
for any one node. So, how do we maintain order? Is there
a market mechanism that can be employed?
That is the very question being asked today by one of our
Senior HP Fellows, Bernardo Huberman. He’s developed
a futuristic piece of software we call Tycoon, which turns
computing grid resources into a kind of stock market. As
a user, you start by opening a bank account and getting credits.
As a user, you place bids for various resources. It’s
proportional, so if you bid $10 and another user bids $5,
you get two-thirds of the resource and the second user gets
one-third. If your deadline becomes more urgent, you can
bid more. The more you use, the more the software deducts
credits from your account.
Bernardo’s team tried Tycoon on a cluster of Linux
servers distributed between our Labs in Palo Alto and the
UK. It did well, and we’ve now given it to CERN, the
European research lab best known to people in this room as
the original home of the Web.
Reinventing the GUI
Another project we have underway is being led by computing
legend Alan Kay, who I mentioned earlier. Thirty-five years
ago, Alan was one of the original researchers at Xerox PARC,
a group of researchers that came together to imagine and
then create much of the personal computing universe as we
know it today.
What is clear is that not a whole lot has changed in those
30 years. The key concepts of a desktop, overlapping windows,
icons, were all developed and then built upon layer by layer
over the past three decades – but haven’t changed
fundamentally since then, even though a lot of other things
First, there have been enormous changes in the power of
computing with high performance networks and processors,
and dramatic increases in memory and storage . . . and second,
even more important, is the changing role of computers. Today,
our computers are in constant communication. We use technology
much more socially than we ever imagined, and the sharing
of information is much more complicated. How could it be
easier? How could it be more efficient? If we were going
to create the computing universe again knowing all we know
today, what would the system look like?
That is the question Alan and his team are working on today.
The team has come up with new software called Croquet. It’s
a new operating system, has a new user interface, and is
a new communications system all at once. The prototype is
in development today. Stay tuned. This software creates an
amazing virtual world -- the closest thing you’ll ever
see today to scalable virtual reality.
As Alan likes to say, “in this world math wins”.
What he means by that is this environment is built on a peer-to-peer
architecture and the more peers that join, the more scalable
and richer an experience it becomes.
Industries reshaped by IT
In the end, the word software is a lot like the word, “electronics.” It
means a lot of things to different people. For us it is something
that is fundamental to all of HP’s interests; and it
is a tool that will continue to change business and that
will continue to change the way we live.
We see the future continuing to evolve with open, service-oriented
architectures, grid and utility service models as well as
newly emerging technologies like the Semantic Web
IT will require management solutions that enable companies
to improve the effectiveness of their IT investments by automating
and virtualizing IT operations. And it will rely on the strengths
of both commercial and open source software,
This view of the future -- for general IT infrastructure,
for services, and for industry-specific solutions -- is what
is shaping the projects we have at HP Labs. And HP’s
increasing focus is on developing both our existing and new
It is an exciting time for all of us in this business, as
industry after industry is being reshaped by software technologies.
We look forward to working with quite a number of you on
hleping to reinvent these industries.