In October 2008, Hewlett-Packard acquired LeftHand Networks, a pioneer in clustered iSCSI storage. In 2009, HP began integrating the technology into its enterprise server and virtualization portfolio, including high-visibility demonstrations intended to show real-time failover between geographically separated storage sites.
I served as lab and deployment support for these demonstrations, working at the intersection of hardware assembly, system configuration, logistics, and live-event readiness.
In parallel, I was also a scheduled speaker at VMworld 2009, presenting on HP Flex10 networking technology integrated with VMware vSphere, which placed me directly within the same enterprise virtualization and networking stack being demonstrated.
For VMworld 2009 (August 31 – September 3), we built a Site A / Site B clustered storage environment in HP’s Houston lab.
The demonstration itself was created by a remote coworker working with a LeftHand engineer. During the Day 1 keynote, a LeftHand engineer delivered the demo: a live video stream that continued playing while workloads failed over between sites.
The demo ran successfully in front of thousands of attendees.
Following VMworld, Mark Potter, then Senior Vice President and General Manager of Industry Standard Servers at Hewlett-Packard, requested that the demonstration be recreated for an already scheduled internal HP sales event at the newly opened Cowboys Stadium in Arlington, Texas.
The original VMworld hardware was immediately redeployed to a demo center after the show. As a result, the environment needed to be rebuilt from scratch on short notice.
For the Arlington event, I was responsible for rebuilding the environment independently.
This second build differed significantly from VMworld.
After initial assembly, a LeftHand engineer joined to support final verification, and I also had remote phone support from the coworker who authored the original VMworld demo logic.
In addition, I worked directly with Mark Potter to walk him through the presentation flow, explain expected failover behavior, and help him present the demo clearly and confidently to a sales audience.
The environment was staged just outside a medium-sized presentation room inside Cowboys Stadium. Due to the temporary nature of the event, the setup consisted of portable equipment rather than rack-mounted infrastructure.
The system operated normally throughout the day.
That evening, additional failover testing was performed.
Although two 30-amp power drops had been requested, the actual setup consisted of 15-amp extension-cord reels connected to standard outlets. During repeated failover cycles, multiple servers rebooted simultaneously, and the combined startup surge exceeded what the cords could safely support.
I heard electrical crackling and immediately observed flames coming from an extension-cord reel concealed beneath a draped table.
I disconnected power immediately.
The fire extinguished at once.
The only damage was the cord reel itself.
At that moment, Bill Haggard, Chief Technology Officer of the Dallas Cowboys, entered the room carrying a fire extinguisher, prepared to respond. Stadium technical staff assisted in redistributing the electrical load using appropriate cabling.
During the power incident, the clustered environment behaved exactly as designed.
The system demonstrated resilience not in a controlled lab scenario, but under genuine real-world stress.
The following morning, Mark Potter and Bill Haggard co-presented the demonstration to approximately 30–40 HP sales staff.
The demo ran cleanly, calmly, and without incident.
Afterward, members of my management chain described the outcome as having “pulled a rabbit out of a hat.”
Mark later presented me with a Dallas Cowboys hoodie from the stadium gift shop — a small but memorable acknowledgment after a high-pressure event.