Google is making it easier for enterprises to do application load tests at scale using the cloud as test infrastructure.
Starting this week, organizations will be able to use Google Cloud Platform, application containers and the company’s Kubernetes container cluster management software to test how their infrastructure will hold up under different application workloads.
The goal is to eliminate the need for organizations to invest in expensive hardware and network infrastructure for conducting load tests, Sandeep Parikh, a Google solution architect said in a post on the company’s cloud platform blog.
“When load testing your application, you first need to setup and provision compute instances, then deploy and execute load testing software,” Parikh said. While the approach can help organizations gain a baseline understanding of their system needs, a dedicated test infrastructure has limited scalability compared to the cloud, according to Parikh.
In order to really evaluate how an application and the underlying infrastructure will perform at scale, organizations typically have to keep provisioning new hardware and executing new tests each time for different workloads, he explained.
Containers and Google’s Kubernetes technology, Parikh asserts, can reduce some of this complexity. Containers can be rapidly scaled, making them a great choice for simulating system clients, he noted.
Cloud platforms also provide a high degree of elasticity, making it relatively easy for businesses to test applications and services with large numbers of simulated clients. “Containers are an excellent abstraction for running test clients because they are lightweight, simple to deploy, immediately available, and well-suited to singular tasks,” Google said in a technical paper describing its new offering.
The paper describes how developers can use Google’s Container Engine to deploy a load-testing framework using multiple containers. According to the company, Google Container Engine allows developers to specify the number of container nodes they need to test a specific workload, the number of users to simulate in a test, and the rate at which users should be spawned.
Developers will be able to see how their test infrastructure performs as the load changes and to measure metrics like the response latency, response failure rates, types of errors and other issues that start cropping up as the workload increases.
The paper itself shows developers only how to test what Google described as a simple Web application. But organizations can use the same approach to create and run much more complex application workloads, the company said.
For example, organizations can do distributed load tests for message applications, data stream management system and database systems. Developers will also be able to customize the metrics that are collected during the load tests and use technologies like Google Cloud Monitoring for reviewing performance metrics, the company said.
Google has not released any pricing information for the new service. Instead, the company pointed developers to its Google Cloud Platform pricing calculator where they can get estimates based on characteristics like the size and scalability of their container cluster test bed.
Google has posted a tutorial on GitHub with step-by-step instructions for enterprises interested in testing their application workload in the cloud.