So, in the previous section, we discussed replication. To prevent failures, our clinic always has two doctors available. Maybe the second doctor served some patients, or maybe they were just patiently waiting for the first doctor to leave for their football game to start working.

But, what would happen to that resilient system if suddenly there is a flu epidemic or a band of rabid squirrels starts attacking citizens in the nearby park? Two doctors won't be able to handle all of the patients and then, again, we have a problem with resiliency. 

But what if we had a supply of retired doctors sitting in their homes playing mahjong? Certainly, we could call them to come and help bandage all of those squirrel victims. And after they were all properly treated, the doctors could return to their mahjong.

That's a system being elastic depending on the workload. 

Elasticity builds on scalability. We could treat all of those patients because each doctor could work independently. But what if all of the bandages were stored in a single box? Then it would create a bottleneck, with all of those doctors standing around waiting for the next pack of bandages.