Scalability vs Elasticity in 99 seconds

Описание: Welcome to a 99-second-tutorial by Zighsys Technologies, on Scaling Versus Elasticity.

As an example, let us imagine an application, application A, running on one server.

With one user, performance is excellent. However, as the number of users increases, performance degrades, creating problems for the users.

The first solution is to get a bigger server with more CPUs and RAM. This hardware upgrade to improve application performance is called vertical scaling.

The second solution is to run the application on 4 servers, and network them to operate as one single unit. This process is called horizontal scaling.

Scalability is defined as the ability to increase workload by adding resources.

Another common problem, is that the number of users varies by the time of the day. Early morning has 2 users, noon 30, evening 15. And by night, the number of users drops to 0.

One solution is to have servers required to optimally support 30 users, 24 7. This is very wasteful.

What if we had the ability to spin different numbers of servers, to match the different numbers of users, during each hour of the day?

When there are more users, we auto-detect that, and we spin up more servers. When there are fewer users, we auto-detect that as well, and we spin down the unused servers. This is possible with cloud services today.

Elasticity can be defined as the ability to match resources, sufficient, for a given workload size.

To summarize, scalability increases resources to keep customers happy; elasticity matches resources with demand to save you money.