Performance Testing Powered with Multi-Region Cloud-Based Load Generation

Public cloud is an extensive platform driving Digital Transformation in industry today. Not only enterprises, but also public sector is open to the adoption of cloud services. The elasticity and scalability characteristics of public cloud platform have won over the industry “Ops” team. With a traditional approach for procurement, lead times have always been a blocker for agile business development. Public Cloud pushes all boundaries to provide the platform for enterprises to reach the market faster. The core business of the enterprise grows without having to worry about external IT delays. 

The Software Development Life Cycle (SDLC) or Application Lifecycle Management (ALM) has four main phases – Planning, Development, Testing and Release. The Requirement Gathering process comes in the Planning Phase. The Development Phase is when the developers develop the deploy-able code package. In the Testing Phase, the quality, performance, or reliability of the application is evaluated before deploying it in production. The last phase is the Release Phase when the deploy-able code package is released into production to the end users, marking the end of SDLC or ALM. Although testing is crucial to determine the overall performance of an application, it demands more effort from the IT team. While evaluating the performance of an application, user traffic and latency are key parameters. To provision a performance or load test, load generation is required. Its puts to test the response time, reliability, resource usage and scalability of an application. The simplest type of Load Test on an application, is URL based Test and the quickest method to provision the load generators for performance testing is by leveraging public cloud. In this way, there is no overhead in the form of lead time or management and maintenance of infrastructure to support performance testing.

Whether the application is hosted on premises or in a different cloud platform, load generators can establish a secure connection to the application behind the firewall, using point-to-site or site-to-site tunneling (VPN Connection). To support multi-region load generation, a VPN tunnel is established between the virtual network in one location and the application hosting premises (shown in figure 1). This virtual network location where the VPN gateway resides, becomes the hub. Peering is formed between the hub and the other geo-location networks. A two-way peering is established between the hub and each of the other networks. The connection from the hub to another geo-location virtual network requires gateway transit permission to send incoming traffic from the application into the geo-location virtual network. Similarly, the connection from a geo-location virtual network to the hub requires remote gateway permission to send outgoing traffic to the application. Figure 1 illustrates the architecture to provision load across multiple regions for performance testing of an application.

Figure 1: Multi-region Cloud-based Performance Testing

The public cloud platform enables us to leverage its scale and elasticity to create quicker releases with faster time to market. Multitude of services are offered across IaaS, PaaS and SaaS to enable the industrial Digital Transformation. Cloud-based load generation for performance testing is a simple use case with coherent industry impact. As the popularity of public cloud reaches new heights, there is a need for every one of us to become comfortable with un-learning and re-learning.