Ox Carts or Dog Sleds
A farmer has to ship 1000 pounds of potatoes to the market 5 miles away. He has a cart and he can use:
- An ox rated 1Hp
- A pack of 12 sled dogs with a total power output of 750 Watts (1Hp)
- 60 Barn Cats generating a whopping 180 calories per second (1Hp)
Which technology should he choose?
Here are a few thoughts:
- For the 1000lb of potatoes the ox would be ideal. It provides brute, scalable (2 Oxen = 2Hp) pulling power
- Moving 200lb a distance of 1000 miles fast, the dogs would do well and scale horizontally (2 packs = 400lb)
- To transfer small individual packages of potatoes to the top of a tree, cats would do much better than oxen
- Mixing the animals seems like a bad idea, we would have some serious integration problems
This is how all this translates to IT:
- A large db server and attached SAN would be ideal to process 10000 purchase orders per hour
- A cluster of commodity servers with distributed storage and scheduler would excel in Monte Carlo simulations
- Desktops are great to run various MS office products enhancing individual efficiency
- Smart phones are ideal to consume small bits of information
The problem arises, when information technologies are forced to serve odd purposes such as:
- Corporate email functions as a document management system
- Ticketing system is used as a records management solution
- Excel-based “integrated” business processes
- People’s memory used for process documentation
- OLTP is used for Business Intelligence
- A public search engine used as your enterprise search solution
- Doodle is the solution for enterprise wide calendar integration
- Building resource loaded schedules in excel
- Use a router as a firewall
The next logical question is: how do you know, that you are using herds of cats to drag your 1000lb of potatoes to the market?
By applying the Watchmaker and Occam’s Razor principles, we can get closer to identifying the problems. If something is complex and difficult, it is not necessarily by intelligent design and may be converted to something simpler.
But it is also good to remember a thought from Henry Louis Mencken: “For every problem there is a solution which is simple, clean and wrong”
At the end of the day it is all about economics. If your interns can enter data cheaper than an automated character recognition system, so be it.
On Monday, sort your customer facing and internal services by ROI and investigate the low ROI items – those that provide little value for a high capex + opex. Then check the under-performers and see whether they are necessary and if they could be done better. Take the maxim of “re-use what you have” with a certain level of skepticism.