Itanium inventor bobs to surface as chip's savior?

Exclusive Some start-ups are comprised of wide-eyed wheelers and dealers with little technical expertise. Others have a decent mix of business types and technology talents. Then there are start-ups like Secure64 Software Corp. that have nothing but the richest pedigree of pure, unadulterated genius running through their giddy veins.

The discovery of Secure64 happened by chance. The company's CEO Peter Cranstone took exception with one of The Register's Itanium bashing articles and sent an e-mail extolling the possible virtues of the chip. This e-mail led to a brief look at Secure64's management team website at which point jaws dropped and little hamsters started turning in heads.

Without slighting other members of the Secure64 team, we have to admit that one name in particular caught our attention - Bill Worley, the startup's CTO. Worley worked on a couple of minor projects during his lengthy tenure at HP. Little things like being the principal architect of the PA-RISC processor and later the principal architect of PA-WideWord - known today as Itanium. Worley, however, didn't just do the initial Itanium designs, he also led the decision, in 1993, to unite HP and Intel behind the project. High-end computing has never been the same since - for better or for worse. And few engineers have a more impressive resume.

Along with Worley, Secure64 has Cranstone, who co-developed the mod_gzip data compression technology for the Apache web server. Its Chariman is Denny Georg, former CTO of various parts of HP. Its VP of Product Delivery is Joe Gersch who once managed HP's research and development organization.

But, as they say late at night, that's not all.

Larry Hambly, one of the first 100 employees at Sun Microsystems, also sits on Secure64's advisory board along with Rajiv Gupta - the former GM of HP's e-Speak web services unit and former head of the joint HP/Intel Itanium development team.

Just an inconsequential, revolutionary OS

So what unambitious project are all these brains working on? Well, just the creation of an abstracted type of operating system that could create faster, more stable, more secure servers.

At present, Secure64 has declined requests for interviews with CEO Cranstone saying the company will have a formal launch early next year. This makes it a bit difficult to know exactly what the company is up to. Thankfully, Worley has applied for a couple of patents that give a decent idea of the direction Secure64 is taking.

At the heart of Worley's recent work is the notion that general purpose operating systems such as Unix, Linux and Windows don't make the best use of specific features in processors - namely features in Itanium. The general purpose nature of today's server market means that systems perform well on a wide-variety of applications, but the boxes aren't tuned as well as they could be for specific tasks.

In the past, any number of companies have taken a stab at this problem by creating server appliances designed to handle a small subset of applications. Most of these appliances relied on sophisticated software to make them different from the average server. Of late, other companies have been trying to tackle the general purposeness of servers with various add-ons. Products such as TCP/IP and SSL accelerators have arrived to speed up the performance of boxes in specific areas.

The appliances and accelerators have largely been aimed at web edge types of workloads - things like serving up web pages, processing web services protocols and encryption. While load balancers and some security appliances have been picked up a decent rate, most of these types of products really haven't enjoyed much interest.

The boys at Secure64 appear to think they've figured out a way to make a web edge system more attractive to customers.