Let's Be Frank
Published: December 18, 2006
by Dan Burger
We can't all be Frank Soltis, IBM's chief scientist for the System i. It's a shame there aren't a few more Franks to go around, though. Makes me wonder whether Big Blue has thought about having him cloned. His engineering accomplishments inside Fortress Rochester have made him a legend. He's the personification of the AS/400, despite the name changes. And he's the most important spokesperson the platform has--regardless of whether you call it the AS/400, the iSeries, or the System i.
The AS/400 faithful love Frank. His sessions at COMMON are standing room only. Same thing when he speaks at local user group meetings. He has enough experiences in the technology field to seemingly fill an L3 cache all by himself, but don't get the idea he lives on past glory. Looking ahead is much more interesting to him, and that was mostly the direction of this conversation I had with him last week, during which we talked about future technologies, new business opportunities for the System i, and cross-platform hardware and software commonalities.
Dan Burger: Before we start discussing the future, let's talk a little bit about what you have been up to recently.
Frank Soltis: One of the things I've been doing--in fact in the past few years I've been doing far more of it--is really promotion. I'm speaking to groups about some of the things we can do with the System i and why it is as good of a business system as it is.
The last trip I was on involved a meeting at a banking technology seminar. Banking executives from all over Asia were in Singapore for this event. They were looking at future technologies that apply to banking and I was telling them about the technologies available with the System i. Some were System i customers, while others were not.
It's good because the System i customers can talk to the others about what they are doing. There are more banks in the world using System i than any other system. That's more than 16,000 banks that use our system. These are individual entities, not a total that includes each bank's branch locations. There are some absolutely huge banks using System i.
When I am in Asia, talking with these bankers, they are talking in terms of having 100 billion accounts and growing to between 150 and 200 billion accounts. It's mind-boggling.
DB: Do you see banking as a growth area for System i?
FS: There is growth for System i. I see a lot of consolidation in this industry. Banking has undergone huge changes. A lot of the older banks have had to upgrade with application changes that add new function and services for their customers. Often the platform they are on doesn't lend itself to doing that and they are looking for a new platform. The fact that the System i is the leading platform puts us in a position for growth.
Another of the reasons I see it as a growth opportunity is because IBM has some very strong business partners in this area. Some of these are new partners that were formerly with other system providers. Some of these other system providers are companies that had been traditionally strong in the banking industry. Unisys, for example, changed its focus over the years and some of its partners that had developed applications for it have been looking for a new system partner. We have teamed up with several of those over the past two or three years. They bring in new business for System i.
DB: Looking ahead to 2007, will you be doing more promotional work?
FS: Probably. First of all, I enjoy doing it. Secondly, a lot of the technical aspects of System i are pretty much in place. I have spent a lot of time working with the Power technology architecture and getting us moved to a common hardware base. Most of that is behind us now as far as the work I do on the architecture.
What I think is important now is getting the word out about the System i and what it can do. I am especially interested in dealing with some new businesses. One that I am looking forward to is the airline industry. We haven't been very strong in that industry, but we are getting interest and we have some strong work going on in that area.
The key to this is the airline industry's expansion. Some of these carriers are growing and they want a platform that can expand with them. A lot of the applications these growing companies use don't scale. The companies are looking for new applications and new application providers.
DB: In your promotional role as a spokesman for the System i, do you deal with companies that are unfamiliar with the System i?
FS: I like to talk with companies that have never seen us before. It's one of my goals to talk with these companies. We need to expand the base and get new customers. We are getting some, but most would agree we are not getting as many as we like to get.
One of the opportunities I see is with companies that are moving away from competitors--particularly Hewlett-Packard. I have met the CIOs of major international corporations that have been strong HP customers and I see two reasons why they are unhappy with HP. Companies are growing very fast, but their data centers are maxed out. CIOs are under pressure to expand, but they don't have any more capacity. There isn't enough electricity, they can't keep them cool, and so forth. One CIO I talked with had over 500 HP PA-RISC servers and over 1,100 X86 severs. This customer is moving to a combination of System i and System p.
The reasons for the change included the capability for partitioning, the capability to run AIX, and the fact that the core business apps at this company were already running on an iSeries that had already proved it could run and run and run. The problems at this company were coming from the PCs and servers that required all the attention.
He also said HP was telling him that he would have to get rid of the PA-RISC and go to Itanium. His decision was influenced by the idea that as long as he had to make that change to Itanium, which would impact all his software, he wanted the best processor technology out there--and that's Power. He told me he was not convinced that five or ten years from now Itanium will exist.
We have a couple of very big, very well-known customers that are doing the same thing.
DB: When you talk about consolidating hundreds of RISC servers and a lot of PCs and giving the System i a bigger role, this is not something that seems to happen all that often. Why hasn't partitioning made much of an impact?
FS: I'm giving you a personal opinion on this, but the way I look at it is that the partitioning capability is there to attract new customers. If you drop into System i shops--or traditional AS/400 shops--those shops probably will have little interest in partitioning. Although Windows will be their second system and we see some consolidation, it is not a real large number.
The same sort of thing goes for customers moving to Linux. There are some parts of the world where Linux is very strong and we see a heavy push to Linux. That is where partitioning support for Linux fits in nicely.
But, for example, I can walk into an HP customer that is running HP-UX, and applications specifically tailored for HP-UX. Those applications will run on AIX. When those customers are forced to go to Itanium, they have to change and making the change to AIX is no big deal. A lot of those customers are running core business applications on iSeries. That fits into the partitioning environment. Our traditional customer that fits into the center of the System i customer base doesn't have a huge interest.
DB: Why wouldn't that customer just put those HP-UX applications on a pSeries box if the move to Itanium is the alternative and he doesn't want to do that?
FS: We can walk into a customer and ask which boxes are giving you problems. If it isn't Windows boxes, it is Unix boxes. Especially with some of the heat generation that is coming from servers today. It makes a lot of sense to use partitioning and put them on a System i. A lot of customers will use a combination of System i and System p. The beauty of that is that things can be moved around. If something is running on System p, it can be moved to System i. It allows consolidation to fewer boxes, but bigger boxes. We are talking about i5 595s and i5 570s.
If a huge data center replaces 500 Unix servers, for instance, it is greatly reducing the amount of energy it consumes and providing more space for expansion.
DB: OK, let's go from Frank Soltis the promoter to Frank Soltis the prognosticator. If you had a clean sheet of paper to work on the next AS/400, what would you be thinking about?
FS: Ha-ha! There are a lot of things I would do if we could start over. That's true of any project.
One of the driving motivations I've had over the past several years is get us on--and I hate to use the term "commodity hardware"--common software. And therefore we could have some common software technology.
If you think about the difference between a System i and an AS/400, the AS/400 was all Rochester technology. Those were the days at IBM where every location did their own thing. One of the things that Tom Watson, IBM's founder, believed in was internal competition. He had all the development labs in competition with one another. When he saw a need for a new product, he had at least two laboratories competing to meet that market need. That kind of structure leads to development organizations not cooperating with one another.
We have joked about Fortress Rochester because, although we got along very well with business partners and customers, but we shut out the rest of IBM. We didn't share our technology and they didn't share their technology. If you go back as recently as the early 1990s, there was nothing in common on any of our platforms. It became clear to many to many of us that we couldn't continue that way. The process, the development, can't be maintained as separate businesses.
I was one of those behind the push to move to common technology. My motivation was not just to be on common technology within the corporation, but I also wanted to be on some common technology outside the corporation. We are seeing that with some of the Power technology. I can't go out and buy a System i from someone else, but in the future I would love to see that.
I believe there is still a possibility that could happen because as we start to expand from this technology and start to move outside into fields that include game consoles, medical imaging, and supercomputers, this technology will get more and more expensive to develop. I am expecting other computer companies will begin buying from IBM, much like they buy from Intel. There will be more systems out there that will be capable of running i5/OS.
Another reason for moving to common technology is our ability to share things. In the next couple of years, you will see more of the mainframe moving to common hardware with other platforms. I have to be careful when I mention this because the mainframe uses a different processor design, but the rest of the technology that surrounds it can be pretty much the same. I expect to see more sharing between mainframe software across platforms.
We also need to share technology for expansion and growth. Certainly the SMB [small and medium businesses] is our heritage with the AS/400. But we are also dealing with huge international companies that want the biggest thing we make. The technology that we had going back in the 1990s was not going to allow us to get there--especially when you look at Power technology and hypervisor capabilities where we could do partitioning and share virtual devices.
I would like to see a more universal product--a System i that would run anything. I've spoken publicly for several years about running Windows on Power. To a limited extent it runs there today. The xBox 360 runs Windows on Power, but that doesn't mean you can take that implementation of Windows and use it as a business application. But I hold out the possibility of Microsoft supporting Power architecture. That, obviously, would allow running Windows in partitions on the System i.
DB: The competition in processor technology is resulting in higher performance, but also more power consumption and worries about data centers overheating. What are your views about the future in this regard?
FS: The whole world of silicon technology will have to change. We are near the limits. When will that be? Well, it's not right around the corner because the industry has a lot invested in silicon technology. That's why you see people rushing out to do dual-core and running things slower. It is to preserve silicon technology. There will come a point in time when we won't be able to do that.
DB: So what might be the next technology?
FS: Within IBM, I think most of the research people think it will be carbon nanotube technology. We are doing a lot of work in a lot of areas. When you are looking to the future and you are not quite sure what technology will be there, you invest in all of them. Some pay off and some you never hear about again. One of the real advantages that IBM has--and other companies have this as well--is the resources to invest in all of these technologies.
Carbon nanotube has tremendous advantages when it comes to size and speed. Incredibly small compared to anything we are doing today. Of course, what you really want is something that you are capable of manufacturing. There are a lot of esoteric technologies out there that we haven't figured out how to manufacture. The secret in this industry is not so much coming up with the technology, but in mass producing it. This is about manufacturing technology and being able to mass produce it. Carbon nanotube looks like something that can work.
My guess is that this will take ten years. I firmly believe we will need something to replace silicon in that time frame.
There are certain things we can do--like with the Cell Power-based chip--but whenever someone has put out something with more capacity and more performance it gets eaten up right away. I mean, we use it.
There are some nay-sayers out there who say "things will never get faster." That's nonsense. There's going to be a need for something beyond silicon. We are going to start seeing that need in the next five, six, or seven years. The industry is going to have to deliver something. Within ten years we will be seeing some of these new technologies.
DB: Let's go back to that idea of Windows running natively on the System i. What if Windows never runs natively on System i? Then what?
FS: If we couldn't do that, another option would be to implement Intel technology along side the Power technology, so customers could run anything they wanted. The beauty of this system is that you are not buying hardware. You are buying i5/OS. I view i5/OS as an application environment more than an operating system. It is really a management environment and you will see more of that in the new releases. You will be able to manage not just a System i but anything and it will be done through a single console. One of the things we have talked about in the next release is the management interface that is being developed in Rochester. That kind of interface that allows the management of Unix, mainframe, and Windows.
Today, companies with different platforms have different interfaces to manage these and they also have different staffs. The training is different for each. We are bringing all that together into a single interface that can manage anything. It is based on work that has been done on the System i.
DB: Does that mean the single console management feature will be exclusive to the System i?
FS: No, it will run on all platforms. It's an IBM project, but developed at Rochester because our people have strong backgrounds in that.
But I would like to see a system that would run anything. Whatever the application a customer has, it would run on this system. At that point customers don't have to worry what the hardware technology or system technology is. They would just go out and pick the best applications for their business. Today, you can't quite do that. We are coming closer. You still have to have different platforms and different support and different management interfaces.
DB: It's hard for me to see a one platform world. Can you really get to that?
FS: You have to set a goal. It's more like how close can I get? That's the real question.
DB: Here's a chance for you to put your promoter's hat on again. What suggestions do you have for promoting the System i?
FS: I don't know if I'll ever be able to pull this one off, but I would like to form an organization called "The One Percent Club." It would be made up of customers that have a total IT budget of one percent or less of their total revenue. I can find lots of System i customers that fit into that. I saw one the other day that was two-tenths of one percent.
One of the fun things to do is that when companies are looking at various platforms is to introduce them to someone from a company that is running IT at less than one percent of total revenue. For most companies the percentage is more like 3, 4, or 5. You can demonstrate to executives how they can drop their IT budgets without getting into the "how much does it cost to buy and what is the base price" questions. We also talk about total cost of ownership. No one understands that because it has been so muddied that it makes no sense any more.
A big part of this comes from virtualization capabilities, staffing requirements, self-managing features, and integration, will reduce operational costs.
We get into head-to-head competitions with Windows and Unix vendors, and our business partners get into these all the time, and this is a way for us to win.
DB: What do you think will happen down the road with build your own application development versus buying off the shelf software?
FS: Buying off the shelf software is clearly the trend. It has been since Y2K. There was a huge shift at that time. And that trend has a lot to do with IBM's efforts in working with its System i ISVs and business partners to modernize applications and bring in new ISVs. We see that as key to the future growth. Businesses are going to be selecting more off the shelf applications.
That doesn't mean people will not continue to develop their own applications or modify off the shelf applications.
I see more companies buying the application regardless of the platform it was designed to run on. Picking the best application will come first.
Then the key to it becomes if multiple platforms can run an application, then I--as a CIO--can get down to what it takes to manage this environment. How many people does it take. And that gets back to the One Percent Club.