The Tender Spot
By stern on Jan 03, 2005
Here are two short-lived examples of monetary standards invented to remove the cost of creating useful legal tender. Fractional currency entered circulation due to a shortage of coins post-Civil War; it replaced postage stamps as a means of small-denomination payment. Shortly after that, with silver prices dropping amidst new mining of the precious metal, the government minted the trade dollar. One of the heaviest silver coins ever producted, trade dollars fluctuated between bullion value and face value, often to the detriment of someone receiving a one as payment. Without well-defined standards for the trade dollar's value, it lost favor domestically and eventually lost its status as legal tender.
Points of today's numismatic detour: Items of value that are easily identified and exchanged are fungible assets. Items of value that are hard to value or exchange create cost headaches. Why I care: As CTO of Sun Services, my team is chartered to create new IT infrastructure models that fix the total cost picture -- not just acquisition, but operational cost over long periods of time. Server assets are not fungible -- you can't easily trade a 2-processor web server for half of a 4-processor web server without a serious investment in security, containers, virtual networking, and resource management. But take virtualization technologies (Solaris 10 containers, resource management, Nauticus VLAN creation, Solaris Security Toolkit features), and apply best practices for management, and you start down the path of making server fractions exchangeable. My job is to keep IT infrastructure from becoming the equivalent of the trade dollar -- heavy, of uncertain value, and hard to exchange at a fair rate.
Next on my list is investigating the software deployment models, instrumentation, and the metering engines needed to create the profit model around the demand model. With models like the dollar per CPU-hour, Sun can attack IT infrastructure delivery the same way Pratt & Whitney does jet engine delivery - we supply the thrust at a cost and weight, and bet on our engineering capabilities to assemble the solution so that it's profitable for everyone. When we say "scaling with technology", we're value engineering our ability to deliver a software stack, via a subscription model, that has a specific performance, service level, and capacity. The only way to solve for both Sun and customer interests is to remove operational cost -- which means more automation, more instrumentation, and fewer people. Anything else is adds cost and complexity, violating the eleven word koans and definitely touching a tender spot with those who fund IT.