Horizontal vs. Vertical Scaling: A Comparison
Want to get up and running fast on AWS? Contact us today.
When hosting applications it's not always easy to predict with any amount of certainty what the future will bring. What sort of traffic will your infrastructure have to support? Will it be enough to respond to sudden traffic spikes from a successful marketing campaign or a new product launch?
Your company needs to be able to respond to demands on your infrastructure, and scaling is a big part of that. Fortunately, that's one of the main strengths of cloud computing. But how should you best configure your cloud infrastructure to be ready for anything? Here, we look into horizontal vs vertical scaling to understand which configuration is best to help future-proof your digital products and services.
Horizontal Vs Vertical Scaling: Which One's Right For You?
Let's start our analysis of horizontal vs vertical scaling with a little side-by-side comparison. Understanding the differences and similarities between these two compute scaling models will help give you a clearer idea of which configuration is right for your organization.
What Is Vertical Scaling?
Vertical scaling is the process of adding more resources to an existing system. These resources can include more CPU, RAM, or storage capacity. Vertical scaling is also sometimes referred to as vertical scalability or "scaling up."
A technical definition of vertical scaling is adding increased capabilities to one component of a system. Increased memory or processing power are the most common example of vertical scalability. Increased storage capacity is another very common example of vertical scaling used by many companies.
Vertical scaling is an approach often taken in the case of self-hosted databases or servers hosting monolithic applications that due to their complexity and dependencies, can only be scaled vertically. Another common application of vertical scaling would be purchasing one very powerful computer and using it to host virtual machines.
This approch to scaling is often used by small- and medium-sized businesses whose workloads are running in on-premises datacentres, or who have constraints in their workloads as mentioned above.
What Is Horizontal Scaling?
Horizontal scaling, on the other hand, is a lot like how it sounds. It's an approach centered around increasing the elasticity and scalability of compute infrastructure by adding servers to your network. Horizontal scaling is also sometimes referred to as "horizontal scalability" or "scaling out."
Central to cloud computing is the concept of horizontal scaling. The AWS service EC2 allows you to add and remove virtual machines quickly and easily in response to demand. There are also services built on top of EC2 such as Elastic Kubernetes Service and Elastic Container Service where you can launch containers within scalable clusters of machines.
Horizontal scaling is virtually limitless. The only constraint is the number of servers you can manage and that your account will allow you to launch. When set-up correctly, scaling out can happen effortlessly on the fly. It's the perfect solution for preparing for unexpected eventualities.
Apache Cassandra is an application based on horizontal scalability, for instance. Cassandra can be deployed with hundreds of connectivity nodes spanning multiple data centers. This redundancy means there's no singular point of failure. Cloud-based servers design to scale horizontally are rugged and robust and prepared for virtually anything.
How To Choose Between Horizontal Vs Vertical Scaling
When it comes to technical challenges, there's no single right way to do things. Every business has its own set of factors that influence the chosen solution to a problem.
Still, there are common technical problems that all companies face, and some approaches to cloud-based computing will be more effective in dealing with these common problems than others.
Vertical scaling has several limitations that can make it less-than-ideal. First of all, scaling up has limitations hardwired to its very nature. Moore's Law indicates that physical computing itself can only be expanded so much. Motherboards and microprocessors can only support so many transistors. You will eventually reach a limit in how far you can vertically scale a server, and it will be expensive to reach that limit. Then when you've purchased the hardware or otherwise made commitments, you can't scale down again.
The other main problem with scaling up is it requires your business to have extraordinarily accurate forecasting capabilities to be truly useful. You'd have to know all of the potential spikes in demand and hurdles your business will face ahead of time to make sure you have enough processing power and storage capacity. For example if Zoom relied on vertical scaling when Covid hit, they wouldn't have been able to respond to the massive increase in demand from everyone suddenly working from home and needing to host meetings online. To make matters worse, if they had scaled up eventually to try to get a piece of the pie, they'd be stuck with too much horsepower when people eventually returned to the workplace and stopped having as many meetings online.
You'd have to have all of your financial resources right out of the gate, as well. As much as we'd all love to buy all of the processors, memory, and hard drives up-front to construct the world's most powerful server, many businesses don't have those kinds of resources when they're first starting.
With horizontal scaling, you can add new components to your system as the need arises. You can continually upgrade, as well, as your income stream becomes more stable and dependable if you're running a lean startup. Essentially horizontal scaling allows you to shift you capital expense (CapEx) to monthly operating expense (OpEx).
Another reason that horizontal scaling is preferable to scaling up is it lets your app or digital business be ready for outages. Downtime can cost even small businesses $10,000/hr. That's not even considering the damage to your reputation that might also occur. The setup that makes horizontal scaling possible also makes it easier to redeploy your workloads in a new region if an outage affects your primary region in the cloud.
If you're offering a digital product, service, or app, its usefulness is entirely contingent upon staying online. It doesn't take very long for a customer to get the idea your app doesn't work and abandon it altogether. The digital world is highly competitive in virtually every arena, so there are bound to be numerous competitors who would be more than happy to take your place.
Horizontal scaling lets you be prepared for anything and save both time and money.
Are You Looking To Migrate Your Applications To the Cloud?
The business world is moving at the speed of thought. You've got to be ready for anything at any time and be able to accommodate a global customer base plus employees located all over the world.