What is the focus of HP’s enterprise software business and what direction will you take moving forward?
Blackmore: Within the Enterprise Group we have four business units: business-critical solutions, the server group, enterprise storage — which is the open SAN and the RAID arrays — and software. The combination of those four businesses is about $US16.5 billion and our R&D budget is slightly less than $US1.8 billion. The six-month planning horizon tends to be “How do we meet our financial plan?” but also “How do we make sure the road map and development [follow] through?” The three-year planning horizon, where Shane [Robison is very] involved, is looking a bit longer. Shane [works] across the company so it’s very valuable to me [to know] are we linking into this, that, and the other? Our industry is developing very fast, and when you’re running a business you’re focused on hitting numbers, not necessarily on where the next disruptive technology is coming [from].
Robison: We think about the enterprise business strategy as it impacts multiple business groups. We’ve got these two huge market segments — consumer and enterprise — and the challenge is how to articulate clear strategy to the enterprise market? It turns out that the software strategy is, in some ways, the point of the arrow for the enterprise strategy. We’ve got a very modularised, low-cost approach to putting together a solutions stack. Starting at the bottom, we’ve cleaned up our microprocessor road map and we have converged on a 32-bit and a 64-bit play. At the operating system level we had a lot of operating systems, but our focus is on NT, HP-UX, and Linux on top of our Itanium and IA32 strategy. At the middleware level we’re going to partner with Microsoft, with BEA, with a couple of smaller market-share players — Oracle and Iona and a few others. At the application level we’ll partner with Oracle, with PeopleSoft, with SAP, with Siebel. The business applications are the drivers for the solution stack and, to some extent, for the architecture of the [IP] infrastructure. So partnering with them very closely and being able to qualify and verify and certify those and also wrap services around them is key.
Where we add value is the management layer of the software. We’ve got an incredibly strong platform, and we have a much richer vision for OpenView going forward. We’re extending it to include the notion of understanding business process changes and then using that to drive provisioning and service-level agreements and all that stuff through an automated layer of software, which manages a datacenter or a fully distributed IP infrastructure where most of our customers are global. It allows you to do all of the things that people talk about in terms of adaptive infrastructure and gives you the flexibility to improve the utilisation of the server farms, their storage infrastructure, and their communications infrastructure. Without exception, the CIOs that we talk to say “I’m not buying a lot of new stuff. What I’m trying to do is understand how to make better use of what I already have.” There’s an opportunity in there, there’s a big software play and there’s a big services play.
Does OpenView have the ability to be a framework like SAP’s NetWeaver of Siebel’s UAN?
Robison: Yes. It will be a layered approach, like most software systems. OpenView today is very strong with element management and the management of the components that you know about. We have the plug-in architecture already there. Then we’ll layer on top of that levels of intelligence and understanding about service-level agreements and about the application layer and business process architectures that will drive it. But we’ll use the same basic approach in terms of having the management, dashboard, and architecture in place at the base, and then we’ll just add the intelligence and the plug-ins as we go along. The idea is to get at the business process level and use that information to drive back through the management and optimisation and provisioning software that we already have.
One of the criticisms about the way enterprises handle computing is that it requires very expensive, very well trained people to watch consoles. Sun took a chance on Sun ONE (Open Net Environment). Is managing things on a component level the best way to go?
Robison: We’re introducing more and more automation at higher and higher levels in the stack. What we really want to do is make the level of management transparent to users. But you can’t build the higher level of abstraction unless that level is really robust, and you have the ability to make it transparent by putting this next level of abstraction on it and raising the level of interaction. We went from a world where there were no tools and you had an army of people running around trying to do all this stuff manually. [First] we got to the basic tools and we [then] got some more sophisticated tools, to [the point] where we want to make that whole process fairly transparent and totally automated. Now we’re looking at the business process level. The challenge is you’ve got to have those building blocks in order to do this highest level one. That’s where we have a real advantage, because we come through this evolution with real technology based on OpenView and other things, like Inside Manager or Storage [Solution] products. We can integrate that together and build on top of that platform. This is where Sun’s got a challenge, because they’ve still got a lot of holes to fill.
Blackmore: They’ve got a good concept, but it’s [a question of] how practical [it is]. Our utility datacenter does what everyone is aspiring to do. It allows you to set service-level agreements at an application level and then allows the computer to do virtualisation, plus knowledge of everything that’s going on in one datacenter or lots of datacenters, to transfer the workload dynamically. You have a combination of virtualisation and dynamic workload transfer, so you can get the absolute best utilisation. And it’s [running] today. Obviously we’re extending it, but this is real breakthrough stuff.
What are customers asking for in terms of abstracting, management, and other IT assets? Do they want you to automate this process and have everything underneath transparent?
Blackmore: They all have the same need. The budget’s gone flat in the last two to three years. Seventy percent of their budget typically is keeping the lights on, managing their infrastructure. The balance is meeting the needs of their business users who say “I want the next generation of the CRM” or whatever. So they’re saying “I need help just managing the infrastructure.” We’re only at the beginning of really leveraging digital information, so the need for [servers] and storage is going to escalate. They’re asking for our help in designing and deploying a solution, and they like this best-of-breed approach because they can choose SAP or PeopleSoft or whatever the current best business application is for their particular need. And they want to be able to integrate that on top of a management layer and a middleware layer and operating system and hardware architecture that’s going to be the best ROI but still high technology solution for them. They would like us to work with them to integrate that whole solution based on their specifications, and then certify it and verify that it’s working and stand behind it and be able to support it with services. The second request is “How do I make the best use of that?” And this is where the management stuff comes in.
What are the missing pieces of OpenView’s stack right now?
Robison: There are a lot of things that we’re looking at [doing to] evolve it. I wouldn’t call it missing pieces, I’d say “How do we continue to mature this software stack?” We’re looking at a lot of the security things, we’re looking at new provisioning software, we’re even looking at things like billing software. Things that will add another level to the OpenView management capabilities that will more effectively address some of these higher level business requests. There are all kinds of things that we’re doing in terms of enhancing things like the directory infrastructure, but it is a very modular approach. We can do that as we go along, and we can do it through make, buy, whatever we need to do.