Inside the Mission-Critical Mind

Howdy!  (That's Texan for hello.)  I'm Mitch Shults.  This is my introductory blog here in the server room.  I work for Pauline Nist, GM of the Mission-Critical segment.  I'm Pauline's 'strategy guy', aka 'mission-critical propellerhead'.

A bit of background on me, then some ideas regarding what I plan to discuss in future blogs.  Please comment on this blog to suggest specific topics you'd like to see explored.

I joined Intel in 1992.  18 years later, I'm still here, much to my surprise.

I say that because I started my career in 1982 with Andersen Consulting (now Accenture) as a systems integration consultant.  I spent eight years there, building mainframe- and UNIX-based mission-critical enterprise systems for the usual Global 500 suspects.  I ended up part of the Advanced Systems Group practice, who were the go-to guys at Andersen for all things technical on implementation projects.  (Which was also fairly surprising, since my academic background is economics, not systems - but somehow computers and me just got along.)

I discovered PC's very early, since I had some downtime in the office one day, and a crusty old partner told me to set up his new IBM PC for him.  That was a pretty trivial exercise, but it led to another project that was anything but trivial - figuring out how to replicate the entire development environment for a 1987-vintage mainframe on a network of Intel(r) 386(tm) processor-based PC's.

Believe it or not, that project was successful.  We managed to migrate essentially all of Andersen's software-development environment for their mainframe-based systems (CICS, DB2, IMS, TSO, etc.) to a network of PC's.

That was when I realized that there was more to these PC things than most folks thought.  If PC's could be used to do mainframe development work with productivity levels that greatly exceeded the native mainframe environment of the day, and at vastly lower cost, then perhaps they could be pressed into service for actual production-level systems deployment!  At least that's the idea that occurred to me after that experience.

Andersen wasn't very impressed with such an outrageous idea (this was 1987, after all!).  So I left to join a small startup called Business Systems Group, where I was the CTO.  BSG was one of the early leaders of the so-called 'client server' revolution, which subsequently became the Web revolution we know today (which is still architecturally client/server, by the way).

Out of the blue, Intel called me up in the fall of '91 and asked if I'd come help them figure out what customers were doing with Intel-based systems with all this client/server stuff.  So I joined Intel in 1992 and moved to Portland from Houston (a vast improvement for an outdoors-oriented individual).

It's been quite a ride ever since.

In 1995, I was directly responsible for the server launch of the Pentium(r) Pro(tm) processor - the event which popularized the phrase 'standard high-volume server'.  The Pentium Pro processor wasn't especially powerful by today's standards, but you have to remember that it was fabricated on .8 micron technology, and thus could only accomodate a few million transistors on a single die.  We struggled to deliver the Pentium Pro at 200MHz, and the biggest struggle of all was to provide a 1MB L2 cache, which proved to be a huge win for database workoads.

Today's Xeon(r) 7500-series processors contain over a billion transistors and run 8 hyperthreaded cores at 2.37GHz, accessing 24MB of shared L3 cache.  A single Xeon 7500-series processor provides more than 100 times the processing power of an entire four-socket Pentium Pro processor-based system.  That's what 15 years of Moore's Law-driven advances in semiconductor technology on the world's most advanced manufacturing facilities will get you!

But just being fast is no longer enough - not when the task is mission-critical.  Since the Pentium Pro, Intel has been improving the mission-critical reliability, availability, serviceability, usability and manageability (RASUM) attributes of the Xeon product line with each generation.  And the Xeon 7500 delivers a truly breakthrough level of capability in these areas.  In addition to RASUM, the Xeon 7500 provides the ability to scale a single system to levels that were previously unachievable using Intel Xeon processors.

As a result, the Xeon 7500 processor can be thought of as the first 'no compromises' Intel-compatible processor for any and all forms of mission-critical computing.  I'm proud to have played a role in the planning, development, launch and ramp of this game-changing processor family.

In future postings, I'll get into some key developments in the mission-critical software and hardware arena that are starting to fully realize the capabilities of the Xeon 7500.  I'll describe some of the many approaches that customers have used to deploy Xeon 7500-based platforms at the heart of their largest-scale, most mission-critical environments, and some of the benefits they've received as a result.

We're in a period of exciting change in the mission-critical segment of the computing industry.  A segment that has long been somewhat hidebound and slow to advance is now being driven forward at the pace of the Intel-based new computer industry.

I'm looking forward to engaging with this community as together we explore the specifics of these exciting changes.  Stay Tuned!