Railroads and trains are a special part of U.S. history. I remember visiting Drew Lewis, CEO of Union Pacific in 1992 when their headquarters was in Bethlehem, PA. The hallways were adorned with historic paintings and the atmosphere of adventure and exploration was intoxicating. Railroaders have their own lingo, some of which we still use to this day. For example “deadbeat”. Webster’s says “one who persistently fails to pay his debts or way.” It originated over 100 years ago when railroad workers noticed that loaded cars passing over rail joints made a different sound than empty cars. The empty ones made a “dead beat” and were seen to not be paying their way.
For passenger transportation in the U.S., railroads had their time between the years when significant rails were laid and the advent of the Interstate Highway system and commercial flights. Even with Carl Grey’s introduction of the Streamliner in 1935, the passenger train dwindled in importance with the exception of commuter loads and the Northeast corridor.
In a similar way, the mainframe computer, or “big iron”, has had its hey day between its commercialization in the 1960’s and the advent of the Internet and remote cloud computing. I remember seeing racks of networked Apple IIe’s already outside the trading room at Citibank London in 1979 that were processing and supporting FX trades in a way the mainframes could not be adapted to do quickly enough.
Trains have continued to be important for inflexible, heavy, bulk point to point conveyance even though trucks account for 68% of tonnage now. Mainframes will continue to be miracles of MIPs, FLOPs and TEPs for specific applications. However, even though trains are now monitored by satellite, this does not allow them to leave the tracks. In much the same way, the software, scaling, and development of servers have ushered in an “off the tracks” flexibility and market responsiveness compared with mainframes.
Size still matters for specific uses that don’t need “off the track” flexibility (airlines, defense, banks). However, size is now available with “off the track” flexibility. Google, for example is estimated to have 1-2 million servers in its data centers and guess what new tech talent is focused on.