[PLUG] Neat Bit from 'Linux on Intel'

[guy1656]

http://www-1.ibm.com/linux/news/michos.shtml

Linux on Intel: A Battle IBM will Win

June 2002

Most people now know that the success of Linux is important to IBM, and have 
heard about IBM's $1 billion Linux investment. But inside IBM's Linux push, a 
key battle is being fought on the Intel Architecture. Rich Michos, IBM Vice 
President of Linux Servers, was recently asked why the battle for Linux on 
Intel is so important to IBM, and how IBM plans to win on this strategically 
important architecture. Here's what he had to say.

Question: Rich, thanks for talking to us about IBM® and Intel. Let me begin 
by asking you why the Intel architecture so important to IBM?

Michos: The Intel platform is the fastest-growing hardware platform in the 
server business, and with IA-64 we're beginning to see greater penetration of 
Intel servers into enterprise computing. This comes at a time when Linux has 
moved out of the universities, away from just doing file, print and other 
basic infrastructure work, and into the enterprise. Linux is showing growth 
that's about four times that of Windows®, and its adoption rate far surpasses 
that of any other operating environment.

Over the past few years, we've adopted open standards for many of the things 
that we do inside of IBM. We have done this at the behest of our customers, 
who have an increasing need for integration. Open standards allow people to 
do other things as well. Our ability to use the Internet is largely based on 
the standardization of protocols like TCP/IP and HTML. We see XML allowing 
applications and processes to share data. We see people utilizing Java™ as 
the standard platform for application development. Now, we see in Linux and 
Intel the standardization of operating system services and microprocessors.

We also believe that as the proliferation of Intel architecture and Linux 
continues, Intel/Linux will draw more applications, and increasing the number 
of applications will further propel volumes. All the while, the relative 
performance of Intel/Linux will continue to improve vs. the other 
alternatives. So in the end, Linux on Intel is a big winner.

But it's not enough just to have an Intel/Linux play. We believe shipping 
Intel/Linux high-volume is critical to IBM. As you know, Sun is a major 
competitor of ours and they've challenged us a bit over the past years.

Now, Sun has organized themselves around a low-volume strategy that's built 
on the Solaris operating system and SPARC microprocessor technology. We think 
that eventually, as the number of deployed Intel/Linux systems runs away, Sun 
may begin to struggle to keep a competitive application stack and competitive 
pricing, even in the enterprise. This will not happen to us.

Question: What does the increasing growth of Intel architecture mean to IBM 
and to its platform strategy?

Michos: It means we intend to dominate this Intel space, and use high-volume 
strategies to get them out the door.

We were in this game early, and we've benefited by working very closely with 
Intel. In fact, we've substantially contributed to optimizing Linux on the 
64-bit Intel architecture. And we're leveraging IBM research and the many 
technologies that we've developed over the years, and applying them in the 
Intel server space to give customers compelling reasons to prefer IBM's Intel 
solutions. We don't change the microprocessor. We get the microprocessor from 
Intel, but we make the processor do more.

At the center of this effort is X-Architecture™ on Intel, and Project eLiza™. 
Mainframe-like partitioning and virtualization on Intel to allow one system 
to run different operating systems, or several instances of the same one. 
This eases integration in homogeneous environments, and enables server 
consolidation on larger Intel-based systems. We've also built features into 
our implementations to greatly enhance I/O. And the adaptive, self-healing 
technologies of Project eLiza predict, isolate and manage failures to 
minimize downtime.

On systems like the x440, customers can easily plan for growth. The x440 will 
let a customer scale from a 4-way to 16-way by simply adding and connecting 
4-processor nodes. We recently announced our easy-to-use Linux Web hosting 
appliances that can be deployed in minutes, and host multiple Web sites on a 
single server at a low cost. And with our upcoming Intel-based Blades 
solution, you'll be able to put literally hundreds of processors in a very 
small footprint, all centrally managed by IBM Director. That becomes very 
compelling for hosting Web applications, and for people that are supplying 
e-utility services. It means that they can manage hundreds of the blade 
servers from a single point of control, and at a fraction of the cost of a 
server farm.

We will continue to distinguish ourselves by not only having 
price-competitive blade offerings, but we'll be able to offer world-class 
systems design and microprocessor packaging. As the volumes will go up, we'll 
continue to ride those curves.

Question: These benefits are mostly hardware and architectural 
differentiators, and that's great. But where does Linux come in? How does it 
help IBM, and why should customers prefer our Intel/Linux solutions to those 
of our competitors?

Michos: One of the challenges we've had over the years is that the world of 
application development has not centered around IBM. ISVs have been writing 
their applications on either Windows, or on UNIX®, where Solaris was 
prevalent. So it's become incumbent upon IBM to find a way to motivate 
application developers to port their applications to IBM operating 
environments. Because Linux runs on everybody's technology, including all of 
IBM's eServers, adopting Linux neutralizes the significance of the operating 
system, and helps position IBM as the platform and vendor of choice due to 
IBM's advantages in technology and support.

By developing for Linux, ISVs can spend more money on function and less money 
on porting. It mitigates any advantage that some of our competitors might 
have had in the past. As we mentioned before, high volume will inevitably 
draw more applications, which in turn will drive even higher volume. It's a 
cycle that feeds back on itself.

Another challenge we've faced has been that he who controls the application 
development environment can also do things like control partnerships for tool 
vendors, and influence the target platform. So just as we have supported 
Apache for web servers, Linux as the open standard for operating systems, and 
Java as the standard development language we have brought vendor-independence 
to the Linux and Windows application development space by open-sourcing $40 
million of code to the Eclipse project. This permits any and all ISVs to 
integrate their developer tools into a vendor-neutral platform.

Linux technical enablements like this are important and necessary, but 
enablement by itself hasn't been enough to motivate enterprise Linux 
adoption. CTOs and IT-directors are a very risk-adverse group. So to complete 
the picture, it's been necessary to back up technical enablement with 
dependable, reliable Linux service and support to let our customers know it 
will be safe to take advantages of the value Linux offers.

Our customers mustn't be confused by all the shouting going on in the 
marketplace about service and support. IBM Global Services Organization (IGS) 
has done several things that make IBM stand out. They have Linux principals 
and practices in every geography. They have a complete portfolio of Linux 
offerings. And the ATS (Advanced Technical Support) organization, the 
technical services arm of IGS and their deep Linux competency, is a real 
asset.

IGS has also recently announced an agreement with SuSE that to a great extent 
allows IGS and SuSE to interoperate as one team. We can act as each other's 
contractors, for example. And we have a two-year technology agreement with 
SuSE to tightly fit the SuSE Linux Enterprise Server product to all of IBM 
eServers.

So in many ways, though technology has helped enable our Linux move into the 
enterprise, it's the IBM brand and the solid reputation of IGS that have been 
pushing it out the door and into enterprise deployments.

Question: You mentioned Web appliances. How do you see IBM's new Web 
appliance play stacking up against the Sun Cobalt appliance servers? After 
all, Sun was first to market.

Michos: Sun's acquisition of Cobalt bought them first into the Linux server 
appliance space, but Linux is still marginalized inside Sun because they 
don't really view Linux as core to their business. They know it's a threat, 
but they haven't quite figured out what to do about it and they're sorting it 
all out now, sometimes in public. Consider that every day, for ten years, 
when a Sun rep or reseller woke up in the morning, the story has been Solaris 
on the low end, Solaris in the middle and Solaris on the high end.

Sun will more than likely leverage Linux in competitive situations when they 
have to. In numerous cases we have removed dozens of Sun servers and replaced 
them with one or two IBM machines, running Linux. That has to be very 
frightening for them.

Question: How do we plan to move our Intel solutions into the marketplace?

Michos: There are several ways that we are delivering Linux/Intel. We are 
working to move more IBM Intel-based offerings into existing accounts, 
stepping up direct sales on the Web, leveraging the partner channel to extend 
our reach into the SMB space; and working with Software Group and their 
partners to accelerate xSeries Intel sales.

There is a misconception that we can't compete on price with our Intel 
servers. Not true. Any customers who believe this should contact their IBM 
sales rep and we'll clarify that quickly.

We will be using IBM.com to move Intel as well. Last month, we launched the 
Linux Store at www.ibm.com/linux/store that will eventually host all of IBM's 
Linux offerings. It will offer the entire xSeries server line, and feature 
the top selling xSeries server on the Linux store front page. Some people 
cede low-cost Intel to Dell, but we have now become cost competitive with 
Dell in the Intel space, and have made great strides in our ability to 
deliver through IBM.com to marginalize Dell's perceived advantage in 
fulfillment.

Beyond selling more Intel into the enterprise, and selling Intel/ Linux 
directly on the Web, we're building a value net around Linux on Intel with 
our broad community of channel partners to move Linux servers through the 
channel. Let me explain how this works in practice, by using a current 
example.

One of the partners that we've negotiated a relationship with is SteelEye, a 
provider of enterprise-grade, low cost, high availability clustering, data 
replication and disaster recovery software. We did some joint marketing with 
them, and now we've begun to leverage their network of partners to help sell 
IBM xSeries™ Linux servers.

They're reaching customers we never would have seen. One of our most exciting 
references so far this year is BBDO, the largest advertising agency in the 
world. It turns out we got that account through a SteelEye business partner 
in Germany that we didn't even have a relationship with.

Then SteelEye came to IBM and asked if we would help sign Champion as a 
SteelEye business partner, and so we helped them do that. Of course, that 
helped us too. So the value net continues to expand. That's how this is 
supposed to work. And we're building a team inside of IBM to help Software 
Group and the IBM xSeries business partners to collaborate even better. We're 
introducing them to each other, training them, and educating them on 
cross-incentives available through the channel.

This is adding even more value to IBM partnership. For example, consider what 
we did with ACCPAC. ACCPAC is a leader in accounting software that 
specializes in delivering country-specific accounting solutions. They have 
grafted to IBM because one of their leading competitors, Great Plains, was 
purchased by Microsoft. They've now completely enabled their application on 
Linux, and Software Group has begun driving integrated offerings with ACCPAC, 
IBM middleware, and xSeries.

So the ISV partners are starting to say, "Linux and Intel is everywhere, but 
IBM can help us drive revenue." That's music to our ears. IBM can help these 
partners deliver a truly integrated offering that customers will be confident 
in, because they know their businesses will be backed up by IBM service and 
support.

Question: Thanks again for taking the time to discuss this with us, Rich. Do 
you have any closing thoughts you'd like to share with our readers?

Michos: Yes, I'd like to leave our readers with a vision of what's just 
around the corner, a vision that we believe will drive our Intel demand 
forward even further: Grid computing.

Our support of Grid computing and the goals of the Globus project are 
starting to remove the importance of "place" from computing. If the server 
you use is on the Grid and the work just happens, you don't care so much 
where or what the processor is. But, if you are paying for computing 
services, you'll really care about price, performance, and reliability. And, 
if you are providing computing services for web hosting, as part of some 
integrated collaboration, or as an e-utility; standards, price/performance, 
and reliability are everything.

Linux plays a key role in Grid initiatives. It is standards-based, delivers 
excellent cost of ownership benefits, and provides the flexibility needed for 
application integration. Many of the notable Grid deployments to date feature 
IBM eServers, and you will see and hear more on this as the year unfolds.

As our mainframe-inspired Intel-based implementations continue to deliver 
greater value than that of our competitors, as Intel microprocessor 
X-Architecture continues to improve performance, and as software initiatives 
such as Project eLiza deliver enhanced management capabilities, IBM will 
remain well-positioned as a leader in Grid technologies. We think the value 
that Linux and IBM server technology adds to Intel will give us an edge here, 
especially when compared with competing Intel commodity solutions. So when 
Grid computing begins to drive customer demand forward, we'll be ready to 
help them.

So, as you can see, we are very excited about our strategy going forward, and 
Linux on Intel is a key element as we charge into the future.

Note: All trademarks are the property of their respective holders.