Comment by protocolture

Had a similar game when I was in a weird role. 2 separate lines of business had their own internal IT functions. However, thanks to a weird set of accountability/responsibility we maintained the hardware/platform of the public webserver while they maintained the website.

So we had 2 pots. The meaner pot was internal to our own team, where we would bet on both how many users would connect to the webserver before it crashed, and then what the other team would blame as the fault. It was always ~3200 and it was almost always RAM.

One of us would sit in on their publicity events, and present the other team with live readouts on hardware usage. The server had umpteen processors with eleventy Jigahertz, and all the RAM that could fit in the chassis (~128GB from memory). 3000 odd users would connect simultaneously, RAM usage would spike to 2%, processor usage would spike to 3% and the website would crash. We would cash in on their pot as to the number of successful simultaneous connections. Then we would go back to our team, and cash in on users AND whatever they were blaming.

After which our IT managers would have their monthly duel where ours would send them a quote to build a better website and they would send us a strongly worded email about how they felt the hardware was the bottleneck.