IBM’s hardware revenue may be continually sliding down but Paulo Carvao, VP of the North American portion of its systems and technology group, reassures us that the problem is isolated to Asia Pacific, and specifically China.

IBM execs told analysts on the company’s Q3 earnings call in October that the 22% drop in hardware sales in China was due to changes in the Chinese government’s economic policy. Still, the quarter was last in a string of three-month stretches in which IBM reported declines in overall hardware sales figures.

At the IBM Enterprise conference in Orlando in October, execs of the hardware business sounded optimistic about both the company’s System z mainframe line and its Power chips and servers. The company is also expanding its cloud-services and big data analytics strategies.

We caught up with Carvao at the event to hear his thoughts about IBM’s role in the future of data center hardware.

FOCUS: It’s no secret that IBM’s hardware revenues have fallen tremendously over the past year or so. What is your plan for reversing the trend?

Paulo Carvao: A lot of the issues that we have had are Asia-Pacific-centric, with a lot of that in China. When I look at my business in the US and Canada, it’s performing better than our business in aggregate.

We’re still in a position in which we are gaining share, especially in areas like storage. And at the end of the day, we’re in a moment in which the industry is transforming very rapidly, and we are developing not only platforms but software-defined environments and we are positioning ourselves for the future. I’m confident we’ll be the leaders in the next wave of this new era of computing that is coming.

There is a trend of hardware becoming simpler and cheaper, getting “commoditized,” while the complexity is handled by software. How does IBM see its big expensive machines fit in what seems to be a largely commodity-hardware future?

First, we do not see our hardware as big and expensive. We are investing in software-defined technologies to virtualize compute, storage, network and power capacity in the data center to help clients deal with efficiencies in IT. This can be achieved not only through virtualization but also through economies of scale and utilization of statistical multiplexing.

At the end of the day, I would assert that some of our scale-up architecture, like mainframe and Power, are less expensive than massively-distributed commoditized [Intel-based] x86 architectures or topologies.

Facebook’s Open Compute Project is revolutionizing the way hardware is designed and procured by big IT end users. It’s made space in the market for new competitors, ODMs such as Quanta and Wiwynn, among others. How you view competition like that, and what does it mean for your business?

We believe in and embrace open innovation. What we’re doing with the OpenPower Consortium is a great example of that. We’re opening up the hardware specifications, the microcode specifications, of our Power 8 platform in a consortium that we formed with Google, Mellanox, Tyan and Nvidia.

We’re collaborating to drive innovation that could potentially be consumed by some of these hyperscale data centers and become an alternative option for some, offering what they do today.

What do you hope will come out of the OpenPower Consortium and what is happening there at the moment?

If you take a look at the different players – we have IBM that has technology for the chip set, or the main processor, you have Nvidia, which is a big player in GPUs and can be an alternative for high performance computing and acceleration, you have Tyan, a motherboard provider and also one of the key suppliers of white-label systems for some of these hyperscale data centers, and Mellanox is dealing with some of the connectivity and networking items in this environment.

I foresee an environment in which we are not only going to continue to produce our server line based on these technologies but where we could have others in the industry adopting this technology and with that, not only do we gain in scale but we also tap into industry-wide innovation. With that, everybody wins.

What are your thoughts on the microserver market that is developing – ARM and Avoton server SoCs, and so on?

The reality is that today a lot of hyperscale consumption is focused on x86 architectures and it’s never good for you to be in the hands of one player. I don’t think this is good economically, and it’s also not good from an innovation perspective.

So, the emergence of other players, like the ones that you mentioned, and also what we’re doing with OpenPower, would represent an infusion of creativity and innovation into this segment of the industry.

We certainly see this as an alternative that would drive lower electrical power consumption. [This is] very much in line with what you see [many companies in] the industry try and accomplish.

A lot of technologies enterprises are adopting today were born in the Internet companies’ hyperscale data center environments. The best example would be Apache Hadoop, which came out of work done by Yahoo! and Google engineers. Do you see microservers, built around low-power SoCs, make their way into the enterprise data center some time in the future?

In a sense, today, if you look at your average large enterprise, they have a combination of multiple architectures. They have systems of record that typically deal with online transaction processing that are built around scale-up type of implementations. You have a growing amount of what we call systems of engagement, that provide IT capabilities and power for the front office. In that area, we have much more presence of scale-out architecture.

From that perspective, we can envision a future in which some of this innovation will eventually be embraced by enterprises. But again, I think the real value for an enterprise is in exploiting these hybrid sets of technologies.

IBM is a big supporter of OpenStack, the open-source cloud architecture. Should we expect some OpenStack-based hybrid cloud solutions to come out in the future, integrating IBM hardware and the SoftLayer cloud infrastructure?

We talked about how we’re doing more around OpenStack with Linux for mainframes or z Linux environments. We’re announcing technologies that are based on OpenStack and will exploit OpenStack.

As you develop an open set of APIs and layer based on open architectures, this will allow you to move from your private cloud to a public cloud and burst in and out. And in doing so, take advantage of public cloud providers like SoftLayer.

So, we’re very enthusiastic about the prospects of what we can do with SoftLayer. SoftLayer has a very established presence with born-on-the-web type of companies. A large percentage of all of the applications on your iPhone today are hosted on SoftLayer. About 20% of the DNS servers on all of the Internet are hosted on SoftLayer, so the capability of taking advantage of those technologies and merging them with what you have today in your private cloud environments is very exciting. This is what we believe is the way to go: this hybrid set of capabilities that will allow you to maintain some level of differentiation of internal IT deployments and also taking advantage of public-cloud type of capabilities.

This article first appeared in the 32^nd edition of DatacenterDynamics FOCUSmagazine. Follow the link for a free subscription.