Horizontal vs. Vertical Scalability

Learn about AWS scalability strategies and mechanisms.

We'll cover the following

Scalability, in the cloud computing field, can be simply defined as the ability of a system to increase its workload by accommodating the growth of demand for work. AWS provides both horizontal and vertical scalability. The specific interest includes architects and developers who aim at providing assurance to them that the system developed is functional, low-cost, and capable of meeting the demand of users. This scalability ensures that applications can handle increasing loads smoothly without compromising performance or user experience.

Horizontal scalability

Horizontal scalability, sometimes called scaling out or in, adds more instances or nodes to a pool to handle increased load or remove from it. This is similar to adding new lanes to a road to carry more traffic. It’s pivotal for applications that experience variable workloads, allowing them to maintain performance during peak times without incurring the cost of idle resources during off-peak times.

Press + to interact
Horizontal scaling
Horizontal scaling

In AWS, services such as Amazon EC2 Auto Scaling and Elastic Load Balancing (ELB) practically take care of the horizontal scaling for users.

  • Amazon EC2 Auto Scaling: This automatically adjusts the count of EC2 instances based on the application conditions so that performance levels are maintained. This service will help us make sure we have the right number of EC2s that are handling the load for the application, optimizing both cost and performance.

  • Elastic Load Balancing (ELB): Automatically distributes incoming application or network traffic across multiple targets, such as Amazon EC2 instances, containers, and IP addresses, in multiple Availability Zones. ELB automatically scales the load balancer according to the traffic to the application that varies periodically, hence ensuring the traffic is even among all instances. This distribution optimizes resource use and maximizes application responsiveness.

Vertical scalability

Vertical scalability, the process of scaling up or down, is the aspect of growing the power (CPU, RAM, Storage) of an existing instance or node instead of adding further instances. In other words, it's like upgrading to a more powerful engine which will roll faster and smooth. This approach suits applications with consistent workloads where the demand gradually increases over time.

Press + to interact

That is usually realizable in AWS through vertical scaling—changing the server’s EC2 instance type to be more powerful or modifying other service resources.

  • Changing the instance type of EC2: EC2 allows scaling up the capability of a single instance in order to serve the increasing demand for the scaling of resources significantly. It allows for a quick boost in performance without the complexity of managing additional instances.

  • RDS, Elasticache, and others: Similarly to EC2, a great number of AWS services allow the modification of settings for some other services with increased efficiency, so that they could take on larger loads without adding more service instances. This flexibility in resource management is crucial for maintaining optimal performance as application demands evolve.

Choosing between horizontal and vertical scaling

The choice between horizontal and vertical scaling often depends on several factors, including:

  • Application architecture: Stateless applicationsStateless applications are those that do not save client data generated in one session for use in another session with that client, allowing each request to be processed independently. This architecture simplifies horizontal scaling by enabling any server to handle any request. are often easier to scale horizontally, while stateful applications can sometimes need some refactoring to be rightfully and appropriately horizontally scalable. Understanding the nature of the application can guide the choice of scaling strategy.

  • Cost efficiency: Horizontal scaling can save money when used on a large scale but needs more complex management and monitoring. Although it’s easy to set up and widely available, the benefits decrease as the maximum capacity of each resource becomes less effective.

  • Availability and fault tolerance: Horizontal scaling inherently supports improved availability and fault tolerance since the load is spread over several instances or nodes. It minimizes the effect of a single point of failure, enhancing the liveliness of the application infrastructure.

Scaling options and triggers

Scaling strategies in AWS Auto Scaling are complemented by various options that define how and when instances are adjusted:

  • Maintain scaling: This will maintain the number of instances constant in time, i.e., the number of instances existing at any time would meet the baseline instance count or minimum instance.

  • Manual scaling: Gives the user the freedom to change the instance both below and above the limits given, respectively, thus allowing the user full control of resource allocation.

  • Scheduled scaling: It enables users to schedule various occurrences within a predefined time. This is perfect for controlling predictable fluctuations in demand.

  • Dynamic scaling: It automatically scales with metrics found within AWS CloudWatch to change the number of instances, hence ensuring the efficiency of resources and rapid scaling of instance numbers based on actual application usage.

Get hands-on with 1300+ tech skills courses.