Why Spend on Sizing & Load-Testing on the Cloud? Set up Auto-Scale

Contributed by admin on 16 May 2013

• 
• Tweet
• Get in Touch

Often, product owners and developers ask the question – what should be the size of my infrastructure on the cloud. And often, load tests are prescribed, planned & executed over several days to understand the number of servers required, the load handling capacity and mean-time-to-failure, among other things. While these tests have their use (for eg. to determine potential costs), the elastic nature & auto-scaling capabilities of the cloud bring up a new dimension to the whole aspect of sizing and load testing. And that is - they nearly make sizing an unnecessary exercise.

While you would want to understand the basic needs for your app ‘to run’ on a server i.e. the memory needs, CPU needs, etc., pretty much most of the load can be handled dynamically at run-time. This is especially useful for start-ups or new apps for which there is no traffic data available and predictions can get squashed in a single day. With auto-scale properly setup, the cloud ensures that your app is running at all times and able to serve any additional traffic that comes. Furthermore it will automatically bring down the number of running resources once traffic drops, thus bringing  you the core benefit of the cloud’s pay-per-use model. This is especially useful in today’s age where-in things can go ‘viral’ in a matter of minutes.

So rather than worry about predicting the likely amount of traffic and the number of servers you should run in order to support the predicted load, just setup auto-scale. When it comes to understanding the size of each server needed, start with the smallest possible server that can run your app with good performance and then choose to scale horizontally. The size can be re-adjusted on a later date if required once there is sufficient information about the traffic and once the application has been tuned further. The cloud also offers several other ways to deal with growing traffic such as using scalable shared caches instead of increasing RAM, using multiple read-replicas instead of a bigger DB server, etc. So rather than spending time and money on sizing or load testing, simply setup on the cloud with auto-scale in place and let the cloud do the rest!

Visit us at Neevtech.com to know more about our offerings.

Tags: cloud auto scaling, infrastructure size, load testing, load-testing on the cloud, mean-time-to-failure