Interview with Rob McKeel
Embedded computing is a great business to be in, because the world is becoming increasingly data-centric and everyone is pushing for real time access to critical data. That applies to almost every industry – manufacturing, telecommunications, the military and so on - giving our business very broad scope: there’s an almost endless array of problems looking for the solutions we can help provide. That accelerating need for real time access to data means that the embedded computing business is a very dynamic one, one in which technology is moving as rapidly as it’s moving anywhere – and our role is to bring that technology to market, enabling our customers to bring more capability to their applications. That’s an exciting – and challenging - position to be in. Best of all, we have a large, global, and talented engineering team, and we work with great partners, and they’re all focused on understanding what those technologies can do, turning them into solutions that businesses around the world can leverage. For you, what were the most important achievements for the Embedded Systems business in 2008? The two most important things in our world are customers and technology, so it’s no surprise that our most important achievements in 2008 were focused on those areas. For our customers, we realigned our business to really focus on key segments. For a technology company, that’s not so easy to do – it can seem more intuitive to organize around technologies. Organizationally, that’s a natural thing to do. But those technologies don’t exist in a vacuum – they provide the potential for solutions, and different customers need different solutions. You can have a really amazing technology, but if it doesn’t solve a customer problem, it’s pretty much useless. Now, we’ve aligned around the market segments we serve, so that we have individual businesses that exist just to bring appropriate solutions to their market. It’s already creating a much stronger link between what our customers need from us, and the products we develop. The second significant achievement has been how we’ve been able to maintain – and, in fact, accelerate – the rate at which we’re bringing new technologies to market, and growing the number of new products based on those technologies. Take communications, for example. We now have what we believe is the widest range of Ethernet switches of any company, meaning that our customers can choose precisely the solution they need. We’ve extended our packet processing portfolio so we are ready to help telecommunications companies solve the challenges of a rapidly growing user base demanding faster access to information. If you look at VPX – an embedded computing architecture that is revolutionizing military applications – no other company offers their customers more options. We’ve launched new compute platforms, new I/O options and a whole host of other products. We’re maintaining that momentum in 2009. What do you see as the single biggest challenge for the Embedded Systems business in 2009? One of the characteristics of the embedded computing world in which we operate is the disparity between product and technology introductions on the supply side, and the extended deployment cycle on the demand side. What does that mean? Well, our customers will typically deploy their solutions over a period of years. If you’re a manufacturing company, for example, you’ll expect the new manufacturing line to cope with your demands for the foreseeable future. In the military, deployment is sometimes measurable in decades. Take the B-52 bomber, for example: it’s still flying today – but it was introduced in the 1940s. But on the supply side, technology manufacturers are bringing out new products – and obsoleting old products – every few months. They are driven in many cases by the consumer market, which seems to have an insatiable appetite for new, bright, shiny objects. The deployed lifecycle of a consumer product is often measurable in months rather than years – and certainly not in decades. But that presents a real problem for commercial customers. They need to be able to support and maintain their systems over the long haul. They can’t afford to invest millions of dollars and then start all over again because they can no longer get the spare parts they need, or they can’t upgrade what they have. That’s where we come in. It’s a challenge we’ve faced in our military and aerospace business for some time now, and it’s a challenge we believe we’ve responded to well. Our job is to help them work through the challenges caused by component obsolescence. We do that in a number of ways, and we can tailor our approach to the specific needs of each customer. For example, we’re committed to bringing out new generations of products that are compatible with what went before, but offer greater functionality and performance – meaning that our customers always have a cost-effective route forward into the future. Some customers might want us to guarantee availability of an existing product for a period of years. Or, for other customers, they may want us to acquire a lifetime buy of key components. Everything depends on what the customer needs from us. But that challenge becomes greater as the pace of technology change accelerates – which means that it will be a bigger challenge in 2009 than it was in 2008, and an even bigger challenge in 2010. What industry standards do you believe will be most influential for GE Intelligent Platforms in 2009 and beyond? Ours is a business that has industry standards at its core. Everywhere you look, there are standards developed and maintained by industry bodies such as PICMG or VITA or de facto standards such as the Intel/Windows environment. Why are they important? Simply because they give our customers a competitive marketplace, breadth of choice, support infrastructure, interoperability and so on – all of which helps to reduce lifetime cost of ownership. For us, there are two very high profile standards. It’ll come as no surprise that they’re assuming greater importance because what they bring with them is an unbeatable combination of high performance from small, lightweight boxes. In the military market, that standard is VPX. The real beauty of VPX is that it leverages much of the military’s huge investment in VME over the past 20 or 30 years, yet lets them achieve levels of performance in space-constrained environments like unmanned vehicles that were previously unimaginable. We really believe that VPX is the future for military embedded computing, and it’s a technology that we’re investing in heavily. Outside of defense, the standard that’s most comparable to VPX is MicroTCA, for very similar reasons. It’s an ideal technology not just for high performance in small spaces, but also because it is designed specifically for environments where uptime is of vital concern. It has high availability and easy maintainability built into it. That’s another area where we’re investing a great deal. What are the key technologies for the embedded computing segment? There are so many, it’s difficult to choose just a few. However, in general, processors are a key technology because that’s where the compute power comes from. That might be processors for commercial and military applications from the likes of Freescale and Intel, or it might be the processors from Cavium that power our latest packet processors. We’re seeing two trends. One is towards multi-core – like Intel’s Core2 Duo, which most people are familiar with. In that space, quad core and beyond will soon become commonplace. In the telecommunications space, Cavium is already shipping its 12-core OCTEON processor, and you really need that kind of compute engine for challenging applications like deep packet inspection. The other trend – and it goes counter to the growing need for processing bandwidth – is towards processors that consume less power and dissipate less heat. Consuming less power is a good thing, obviously, for cost and environmental reasons, but in the growing number of portable computing applications, it’s imperative, especially where the application relies on battery power. Then, in many applications – especially the military – processors that consume less power and dissipate less heat are vital because they are increasingly being deployed in small, lightweight subsystems and because cooling is such a challenge in that environment. Processors like Intel’s Atom are responding to those low power needs, and are a good fit as we develop new products that are smaller and lighter and focus our attention on architectures like VPX and MicroTCA that were specifically developed for problems that need compact, lightweight solutions. What trends do you see that the embedded community will need to respond to? Today’s applications need more processing power, and new applications are emerging all the time that are really compute-hungry. Space is at a premium in just about every environment, and our products need to enable solutions that are smaller and more compact. Power consumption is not going to become less of a challenge, and we need to track developments in that area, developing solutions that use less energy. The motto of the Olympics is “faster, higher, stronger.” Our motto needs to be “faster, smaller, lighter.” There are other emerging trends, though, that we need to think through how we respond to. Perhaps most important among those, especially given what I said before about our growing reliance on access to information, is for security of data. That could involve a whole host of technologies including physical security, logical security, anti-tamper protection - something we’re starting to see growing demand for from our military customers - and so on. If there was one thing you could change about the embedded business, what would it be? One of the things that challenges us most is the concept of COTS – commercial off-the-shelf. The thinking behind COTS – and while it applies mostly to the military, it also applies to a large extent across all industries – is that customers will source what amounts to ready-made solutions. By definition, those solutions will be readily available, will leverage leading edge technologies, will be priced competitively and will be relatively easier to sustain and support over the long term. That’s the theory. However, the practice is somewhat different. COTS is a starting point, rather than a destination. Despite the vast number of suppliers and products, many major organizations – many of whom have grown up in an environment where custom designs are the norm - believe they need something just a little bit different. They’ll say: “We really like your XYZ board – but can you make it for us with this functionality that we really need?” It would certainly make our lives easier – and our customers’ lives as well – if everything we supplied was indeed standard, off-the-shelf products. And there’s no doubt: we’ve made huge progress in that direction – as an industry and as a company – over the past 10 and more years. Customers clearly see the benefits of COTS in terms of long-term cost of ownership. But still… So we as a company need to be realistic about the realities of the marketplace – a realism that shows itself in three ways. First, we continue to develop the broadest possible range of platforms so that we have more off-the-shelf starting points. Second, we design those platforms knowing that some customers will need them to be modified – so we design flexibility in from the outset. And third, we are organized to be nimble and responsive, with great engineering talent, so that we can respond to those different needs more rapidly than our competitors. |
Rob McKeel is Vice President, Embedded Systems for GE Intelligent Platforms. The Embedded Systems business – which is part of GE Intelligent Platforms – serves a range of industries, including commercial, industrial, telecommunications, military/ aerospace and avionics. Here, he talks about where the Embedded Systems business has been – and where he sees it going.