Online businesses expect more users and more usage of their apps. Lot of money is spent on marketing to pull crowd. With that in mind, their IT teams get ready to face the peak crowd and provision hardware accordingly. There is no easy logic to calculate the number of users who would visit your site on a crazy deal day. In this process, there is always a question that needs to be answered - are we sizing the right hardware for the right crowd?

The first step is to understand the visitor trend for the last 1 year. If you have that statistics, using a real user monitoring tool, that is a great starting point. Trend in normal week days, weekends, festival season etc., in the past year will help us to project for current and future.

But the increase in visitors will depend on your current marketing spend as well as that of your competitors. At this step you can to some extent associate a certain percentage growth from your past to current to future. But nothing can actually predict how the real users will react.

The next step is to ensure that your app can withstand the projected number of visitors and transactions. If your peak crowd was 10000 visitors in a span of 4 hours in the last festive season and you expect 18000 visitors in 3 hour span this year, get the drill done. There is no other better way than pushing your app to its limits. See that in action with your own eyes and then spend the money on marketing.

To do this, you need to conduct load test on your app. This is not just a one-time exercise, every year. Whenever you do major releases and before any big push marketing campaigns, this needs to be done. Remember 3 things that are going to affect your system's performance - more users, more data and more usage time.

By doing this, you will catch any slowness in your app and preempt the chances of app crashes. If your app performs slower, you must use a deep-dive APM tool to get code level visibility. Even after performance tuning of app, if the app is slower, it may be time to check your hardware sizing.

You can use the power of cloud to have on-demand provisioning of servers using elastic load balancers. This is advisable, if your app is on cloud. For on-prem data centers, to calculate the right size for your hardware, you must conduct a few iterations of load tests with different number of virtual users. For on-prem, it is better to do load tests multiple times, size the correct hardware and then go live. With traditional data centers or on-prem, you will not have the luxury of adding servers after going live.

Appedo can help you to achieve this recorded webinar on load testing