When my kids were little, I sent them to Economics Camp.
I know what you are thinking, "What kind of warped Dad sends his impressionable children to study the dismal science?"
Well, I did, and they loved it.
One of their favorite activities that day was an exercise to show how restraints of trade are bad. If you have ever taken a class with me, you have heard this story, but for the May ACE training in White Plains, I decided to actually do the exercise, rather than tell the story. I had always wanted to do the exercise, but one needs a sizable group of participants to make it worth while and the time to obtain some supplies beforehand. This time, I had both.
I usually tell this story at the beginning of the discussion on groups and policies, because I want our customers to understand that if they want performance and efficiency, the market is the best way to get there, and constraints limit the free market. Limiting the market with constraints, therefore, limit performance and efficiency -- so unless you really need to, don't make constraints. But that day, I was going to give them a visceral experience of the problems with restraints on the market.
With the help of some friends from the COFE team, we provisioned 24 paper bags, each with a small assortment of goodies. There were about 3 - 5 items per bag, each one had a different assortment. Here are some of the things you might have found in your bag, had you been there:
- Bite-sized chocolate bar
- Hard candy, wrapped in cellophane
- Ball point pen
- 2" x 7" sheet of metal foil stickers
- Clear plastic cylinder containing toothpicks
- Clear plastic cylinder containing Q-Tips
- Pack of gum
- Bag of nuts, cajun spiced
- Bag of chocolate 'money'
- Travel-sized pack of tissues
- Chocolate Yoo-Hoo drink box
We handed out the bags to the students and then asked them to rate their satisfaction on a scale of 1 to 10, with 10 being the highest. I asked them "How happy are you with what you got in your sack?", and then I collected their results:
In this first round, a few students were really happy with the contents of their bags, as you can see from the 10's, and the lowest was a 2. The average was about 6.
So then I told the students "You can trade with the other person at your table, if you want to." This gave each student one trading partner. They were tentative at first, but quickly realized an opportunity to improve their stash.
After they were done trading, I again asked them, "Now, how happy are you with what you have." Here were their scores after trading:
Wow! With only one trading partner, average satisfaction went up 15%! Everyone either stayed the same or went up. This is to be expected because nobody trades unless he thinks that what he is getting is worth more than what he is giving up to get it.
For the next round, I told the students they could trade with anyone on their row, which gave each person 5 - 7 trading partners.
They really got into the game at this point, and here are the satisfaction results:
There were a few surprises here: First notice that S2, who started at 7, then went to 10, went down to 9. I asked her why, and she said she became less satisfied, not as a result of bad trades, but because she had come to realize what other goodies were out there (her horizons were broadened by increased trading opportunities). One student, S17, stayed at 7. This person actually was not very interested in the contents of his bag and was only trading to make others happy. Everyone else either stayed the same or went up, and the satisfaction average after trading with the row increased to 7.6, a 27% improvement over no trading.
Finally, I told the students they could trade with anyone in the room, and things got crazy. One student cornered the market on Yoo-Hoos. And there was a bidding war for one of the sticker sheets. One student figured out that the bags were a commodity as well and started trading for bags.
It was sheer pandemonium!
As you have guessed, the satisfaction results were higher still.
Average satisfaction had now risen to 8.1, a 36% improvement over No Trading. The average increase for each person was 2.1, or 21%.
The students had a great time doing this exercise, and we all learned an important truth about economics: when you have the option to trade, you have more potential satisfaction. And more opportunities to trade make for greater and greater potential satisfaction.
It is the same in the datacenter. Every constraint in the market limits the potential options for the VM, and thereby, reduces its potential. We need to unlearn decades of collective wisdom which have led us to sequester resources behind cluster boundaries and tiering policies, and free the VMs to go wherever they want, to get the resources they need.
Thanks to my wife for helping me pick out the goodies.
Thanks to Martha for her Edit-Fu.
Thanks to Martha and Divya for helping fill the bags and distribute them.
Thanks to all the students, Francis, Rebecca, Martha, and Divya for playing along.