The Unix Guardian--Q&A with HP's Brian Cox: Tukwila Itaniums and Hockey Pucks

Newsletters	Subscriptions	Forums	Safari	Store	Career	Media Kit	About Us	Contact	Search	Home

Volume 5, Number 8 -- February 28, 2008

Q&A with HP's Brian Cox: Tukwila Itaniums and Hockey Pucks

Published: February 28, 2008

by Timothy Prickett Morgan

As one of the three remaining commercial-grade Unix platforms, Hewlett-Packard is still very much a believer in Unix even as it has created an Integrity platform that supports HP-UX (but not Digital's Tru64 Unix, alas), Linux, Windows, OpenVMS, and NonStop software stacks. While being cross-platform is important for the Itanium-based Integrity line, there is no doubt that HP-UX remains the dominant platform driving Integrity server sales. And that is probably not going to change any time soon.

HP-UX, which someone in an interview I was doing referred to as "Hockey Pucks," as well as some other funny variants you cannot print in a family newsletter, is one of the key drivers of sales and profits at HP. That's because HP-UX is the central component of a $20 billion-ish ecosystem of hardware, software, and services sold by HP and myriad other partners. Brian Cox, director of software planning and marketing for the company's Business Critical Systems unit, is one of the top HP execs who drive the HP-UX platform and who speak for it as well. Cox took a few moments out of his schedule to talk Unix shop with IT Jungle recently, and business is good.

Timothy Prickett Morgan: I was hoping to talk a bit about the Integrity server line and focus a little bit on Unix, because as you know, I love Unix. I am, as you might imagine, interested in what HP has planned for 2008 and the trends you are trying to ride. It is too damned quiet in this market right now, to put it bluntly, and I am trying to figure out what is selling and why. There is a new processor, the "Tukwila" Itanium that might be coming later this year, and HP undoubtedly has a new chipset and server line based around it.

Brian Cox: We've done this sort of thing before, haven't we? We just closed out 2007, and I can talk a little bit about the trends that we saw this past year, and I can tell you about Unix growth and the drivers behind that.

If I recall the pundits' predictions correctly, 2007 actually turned out to be a pretty good year for Unix. IDC published their last results in November 2007, when we had not quite finished out the year, but the Unix server business grew over 4 percent in the first three quarters of 2007. So that doesn't sound very much like it is dead and buried--much less shrinking--to me. That's the overall Unix server market. Where HP is particularly strong is in the high end of the Unix space, and that grew at over 9 percent year over year in those three quarters, and we are pretty happy about that. Our particular HP-UX server business grew, from top to bottom, grew 8 percent year over year in those quarters as added together. Which means we gained back some market share points.

[Editor's note: This interview occurred several weeks ago, and IDC is getting ready to put out its 4Q 2007 server numbers any minute now. Stay tuned.]

It has been a good market, and we have had a very good close out to our financial results, if you saw those. . . .

TPM: I did indeed.

BC: Our Unix business was up, and basically our overall enterprise server and storage business just blew off the doors. We were really happy, and we see Unix as being a fairly robust opportunity for us. This is a large market--about $19 billion as measured by IDC--and that is great business.

So where are customers embracing Unix? We are getting a lot growth in the emerging economies such as China and India, of course. But you have to realize that in a lot of these countries, they do not have a long legacy of big iron. So now the companies located in these countries are really doing their buildout and they are now making their big iron purchases--and more often than not, they are deploying Unix servers. And apparently some are even buying mainframes. But companies in these regions are buying Unix servers at a much greater clip as they power up very large SAP ERP installations, new banks are coming online, telecommunications continues to grow. And in each case, they need bigger boxes to handle traffic. Our business is great in these areas, and it is a fabulous greenfield opportunity for us.

With regards to the more mature economies in North America and Western Europe, we still see growth, which is being driven by certain technology trends. There are two main trends--the scale out trend where you can buy on an incremental, piecemeal basis. One way to build out would be blades--which have been in the tornado phase, if you use the Geoffrey Moore analogy--and that is helping customers address issues of space, efficiency, and flexibility. The other end of the spectrum is consolidation using bigger boxes. They are using virtualization tools to accomplish this, so you can put many, many server instances on bigger boxes to get rid of the server sprawl. That's one reason why I think HP's high-end server business, and the high-end of the overall Unix market, is up. We're seeing more maturity in virtualization tools, which is enabling the server consolidation.

TPM: What does the Integrity roadmap look like for 2008? What kinds of products are you cooking up? Is the quad-core Tukwila Itanium and the QuickPath interconnect actually going to get out the door this year? My assumption is that Tukwila cannot just plug into the "Arches" chipset used in "Madison" and "Montvale" Itanium 9000 servers.

BC: I can't comment on the timing of Intel's products--that would be crossing the line. But what I can say is that when we do move to the Tukwila platform, we are going to be moving to quad-core technology--Intel has been open about that--and there will be some other enhancements in Tukwila as well that will be quite favorable for us. And at that point, we will be looking to start updating our Integrity server line. Some of our platforms, like Superdome, have been around since 2000. And in order to start embracing the new technology, you are going to see new platforms coming up. As we get closer to launch, I can give you more details. But that is all I can say so far.

TPM: To just be a little more precise, Tukwila chips will not and cannot plug into Arches-based Integrity machines.

BC: We plan to do a new chipset at that point, and HP is not ready to talk about code names quite yet.

TPM: Let's talk a little bit about HP-UX 11i. We went over the roadmap a little bit last year. It seems to me that operating systems are moving in kinda s-l-o-w m-o-t-i-o-n now. Customers seem to want that, and as a data closet--not data center--manager, I get that.

BC: In the enterprise space, when you are running your mission-critical workloads--whether it is transaction processing for a bank, handling 911 calls for the police department, running stock exchanges, billing for telecoms--this is stuff that you just can't screw up. You are not going to rip and replace just because of some hot, new buzzword technology. You want something that is absolutely rock solid, and that is why we believe Unix has been doing really well. It has such attributes, and it does so with a much lower total cost of ownership than what we have seen historically from mainframes.

This is why the emerging economies are not just throwing together a scattershot cluster of Linux boxes trying to run their stock exchanges. They realize what they need to do in terms of quality of service to satisfy their investors. The same holds true in telecom or manufacturing.

What I think you are going to see with Unix is continued enhancements. But you have to realize--and this is the way I always looked at it--Unix is where Linux and Windows want to be. That is their end state in terms of features, robustness, scalability, security, manageability, and so forth. In fact, if you talk to industry analysts, they complain that people think that VMware invented virtualization just because it gets a lot of press.

TPM: No, no, no. VMware invented a means of making a lot of money off of virtualization--that's different. . . .

BC: Well, I think IBM actually invented virtualization for the mainframe.

TPM: Kudos to clone mainframe maker Amdahl who started it, actually.

BC: I went to the Gartner Symposium a few weeks ago, and John Phelps, one of the analysts there, explained that if you looked at the features in virtualization, they are much more robust on Unix systems than they are in products from VMware, Citrix Systems, and others. They are still trying to catch up, and they get a lot of press because it is running on a high volume, X64 product, but their tools are still not as mature as those on Unix and mainframe systems. So that is another reason why consolidation on these big iron boxes is still highly trusted, and that is what customers using these machines for back-office processing are doing as they consolidate.

So in terms of Unix platforms, and HP-UX in particular, we will continue to see enhancements for virtualization. We will be working on ways to better manage all of the virtualized images--how do you inventory, deploy, charge back, and do the mundane systems management tasks that you have to do. If you talk to the industry analysts, they say that HP integrates its virtualization tools best into its platforms and its management tools. While IBM has a pretty robust set of tools, they are just not instrumented to work together as well as what HP has done. Sun Microsystems is even more scattershot. Sun does have some good tools, but there are some big gaps compared to HP and IBM.

We are going to continue to scale up HP-UX as we add processor cores to the servers, and we will be focused on scalability tuning in our Unix. But it is more than that. We're still very good allies with Microsoft, and we are a huge contributor to the Linux community. It is just going to take time. HP-UX is going to have its 25th anniversary this year, and it took us decades to get HP-UX to the place where we are today. Linux and Window just have not been around as long on servers, and it will take them some time to catch up. Probably not 25 years. Some people say that they are good enough to do banking systems and stock exchanges, but that still doesn't mean companies are going to rip and replace the Unix applications that have been working so well. There is a long, long life ahead for Unix. There's just no doubt about it. We have had 25 years with HP-UX, and there will be decades more--I will be retired--before HP-UX fades into the sunset.

TPM: What do you think about IBM's AIX 6.1 and Sun's Solaris 10 and how you are positioned against them? For the past decade, vendors have used hardware pricing as a lever to get customers--a very small percentage of the Unix base, mind you--to switch platforms.

BC: They're solid operating systems. Credit to those companies. They committed to Unix, and they made investments in it and continued to enhance their Unix just like we have.

I think that there tends to be a different emphasis in the different Unix offerings. Classically, Solaris has been really strong with deployments on the edge of the network and Internet, whereas AIX and HP-UX have tended to be more focused on reliability and back end database serving and transaction processing.

I think HP-UX is the Unix that works best in a multi-OS operating environment, because we have made sure that our management tools work across Unix, Windows, and Linux platforms. We have been sincere in our approach to this, because we have a huge X86 and X64 server business that is based on Linux and Windows. We have to do the tools right if we want to maintain our market share.