Scalability Versus Elasticity: What’s the Difference, and Why Does It Matter?

The cloud is growing rapidly as a means to absorb the 8 zettabytes of traffic Cisco expects by 2021. One thing IT pros have been seeking in the cloud is scalability. Or is it elasticity? Or both?

It turns out, one of these features generally attributed to the cloud is, in fact, more “cloudy” than the other. Here’s why.

Definitions First

Scalability and elasticity are often confused, but they are distinct attributes of a data center or cloud environment. Scalability generally refers to more predictable infrastructure expansions. If a particular application gains users, the servers devoted to it can be scaled up or scaled out.

Most organizations reevaluate resource planning at least annually or, during periods of rapid growth, even monthly. As they predict more customers, more employees, etc., they can anticipate IT needs and scale appropriately. This can happen in reverse as well; organizations can downscale in response to business fall-off, increased efficiencies, and other reasons.

Elasticity, on the other hand, is useful for discussing shorter term resource needs, such as sudden bursts of traffic that could threaten to overwhelm an e-commerce site. With elastic infrastructure, workload capacity can increase to fit the surge of customer interest and then “snap back” to regular levels, preferably like the waistband on a new pair of shorts, not those worn-out favorites.

With elasticity built in, IT organizations can resist expensive overprovisioning for “just in case” scenarios and instead draw on—and pay for—those resources only when they’re needed.

On-Premises Versus Public Cloud?

The real difference between scalability and elasticity lies in how dynamic the adaptation. Scalability responds to longer business cycles, such as projected growth. Elasticity can handle the up-and-down nature of website hits, sales demand, and similar business needs in a rapid and often automated manner. Organizations with sudden or cyclical changes will most often need elastic capabilities in at least some areas.

As more and more organizations look to hybrid cloud environments, scalability and elasticity needs can delineate which services belong in a public cloud environment and which can be handled by the enterprise.

The public cloud excels at elasticity. AWS, Microsoft Azure, Google Cloud, or other providers can easily ramp up servers to stream the exciting conclusion to your expensive Superbowl ad. On the other hand, on-premises IT would be inherently less efficient if it had to maintain capacity sitting idle, just in case Dr. Oz were to make mention of your product on air, driving millions of orders overnight.

By the same token, on-premises IT deals very well with low-latency needs. And to date, it’s often the trusted solution for many mission critical applications and those with high security and/or compliance demands (although that’s changing to some degree).

Many ERP systems, for example, need to be scalable but not exceptionally elastic. Running them on owned, not pay-for-use, equipment—even in a virtualized, self-provisioning, and other “cloudy” environment—is often the best answer.

The balance can shift further toward on-premises for the right use cases when IT also controls data center costs, including IT hardware maintenance. When a server or storage array can run five, eight, even ten years or more with strategic upgrades, third party hardware maintenance, and affordable post-EOSL support, it may offer cost-savings over a pay-as-you-grow solution.

As with so many other IT questions, scalability versus elasticity—as well as owned versus rented resources—is a matter of balance. But understanding the difference and the use cases is the starting place for finding the right mix.