Scaling Your Web Applications in the Cloud: Manual vs. Scheduled vs. Automatic

By Sebastian Blicharz| June 27, 2018

The cloud has completely revolutionized the IT realm, opening up exciting new opportunities that make it easier than ever for companies to scale their growth. However, managing a cloud infrastructure means looking at how you make budgetary decisions in a new way. While there are significant economic benefits to cloud computing, having the ability to scale resources up and down can make budget forecasting a significant hurdle for many organizations.

When it comes to cloud spending, the good news is that you can scale up in times of greater demand, and scale back down in seconds once the need is no longer there. This can be a sustainable model, but it will require a new approach to modern IT management.

There are three primary ways to manage web app scaling in the cloud. Let’s break down the pros and cons of each method:


Pros: Monitor app performance by manually collecting feedback from users is a good way to identify when the demand has outgrown your hosting. Based on your analysis, you can scale the app as needed in seconds.

Cons: The downside to manual monitoring is that performance issues must be caught early enough to prevent poor user experiences. If you don’t manually scale down soon enough, you’ll pay more in off-peak hours.

If you’re currently running applications in Azure and are drowning in the manual process, Azure Advisor is a great tool that can help maximize your resources!


Pros: If an increase in demand is predictable for your business case, then scheduling scaling could be a good way to stay ahead of those ebbs and flows. For example, if you know application usage goes up on Friday evenings, you can schedule around that.

Cons: The disadvantage here is that most businesses will inevitably be confronted with unpredicted spikes. When this occurs, users will experience poor performance without a backup plan.


Pros: If upticks happen gradually over the course of 10+ minutes or a few hours, automatic could be the ideal option for you. This method takes into account CPU usage, memory usage, disk usage, number of queued http requests, and other metrics to trigger automatic scaling in either direction based on need.

Cons: To prevent unnecessary scaling, automated systems measure demand over a 5 minute or more duration. Sudden spikes can leave gaps, meaning that users can still have performance issues and application errors.

Before deciding which method to use, you’ve got to understand the root of your performance issues. Adding more CPU stores to the web server won’t help if the issues are related to high memory usage, disk I/O, or database performance. It’s also important to consider optimizing your applications.


Tech Insights Report