Keshava Raju, Head of Solutions, NFV & Cloud Assurance, Spirent Communications
Almost all enterprise customers have deployed and are operating virtualised data centres. They are on the path of transforming and consolidating their virtual data centres into the cloud. Cloudification offers tremendous benefits – greater agility, shorter time-to-market, lower costs, high returns on investment (ROI) and the potential to avoid vendor lock-in. Enterprises choose the right cloud to adopt based on benefits such as ROI, operational efficiency, application performance, resiliency, quality of user experience, etc. However, as business applications migrate to the cloud, reliability and performance become even more critical. It is crucial to test the infrastructure that is moving from on-premises enterprise to the cloud – all with flexibility, stability, and improved deployment and provisioning.
Today, enterprises have a mix of cloud models – public (Infrastructure-as-a-service/platform-as-a-service/software-as-a-service), dedicated, private and hybrid. Many enterprise applications and next-generation services, particularly those that use private 5G/SASE technologies, span a mix of multi-cloud environments – edge clouds, private clouds and increasingly public clouds.
Testing the cloud
If the performance of cloud infrastructure is poor, applications hosted on that infrastructure cannot provide an improved user experience. Application migration may pose several challenges: it can be time-consuming or cause performance issues in the target cloud, which could lead to rollbacks and risk outages.
In order to ensure successful cloud migrations, it is necessary to have a solution capable of validating and ensuring cloud infrastructure performance in multiple dimensions: compute, memory, storage and network. With everything disaggregated, it is important to perform testing at every layer of the cloud; the testing approach must be considered from two aspects: horizontal and vertical.
It is important to test each segment or domain of the multiclouds individually and then assess the end-to-end service performance across segments. Without this visibility, it could take months to launch a service. To perform this testing, cloud test agents (load generators) can be spun-up at key points in each of these segments to test individual functions, before they are put into the live network to make sure that the end-to-end services are working as expected and satisfy better user experience service level agreements (SLAs). Also, security vulnerabilities exist in every area: internet, data centre, software defined-wide area network. Therefore, cloud providers and carriers need to validate the infrastructure of security operations, including functions for inspecting traffic and enforcing policies.
Before running live traffic over these services, wraparound testing techniques should be used to create incidences that will trigger these traps and send information to upstream monitoring and security information and event management systems. The result is end-to-end visibility, including assessments of performance and security concerns.
The vertical aspect is the one that poses a new challenge for cloud providers, as they move towards cloudification. The new world of cloud has layers from different vendors. Testing and assurances isolate the different layers using virtualised and container-based test agents and methodologies. Test agents can surround a virtual/container application to test the function itself, or replace the virtual machine (VM)/container, mimicking the load of the VM/container application to assess the performance of the infrastructure components. Here, too, being able to instantiate test agents helps in understanding if each layer is performing its function well enough to deliver the SLAs needed.
The network security paradigm is changing, too. Networks are shifting to distributed models with localised and optimised security to protect the resource, and all of this is leading to a new attitude towards building the network. Service providers are now adopting continuous integration and continuous delivery (CI/CD) practices, converging labs, and pre-production and production networks.
Automation and analytics
Automation is the next most important aspect of the solution. CI, continuous testing and CD methods enable and deliver the agility that is a key ingredient for success. There needs to be a closed-loop analytics platform that can ingest and understand the performance of the entire cloud network that is delivering the service to the customer. Test platforms leverage machine learning and artificial intelligence techniques to tie together all the network data from various sources, proactively predict service degradations, pinpoint root causes and drive remediation actions. Turning data into actionable insights rapidly reduces the mean time to repair and helps avoid unnecessary escalations. Agile, automated testing and assurance solutions provide a roadmap for leveraging cloud networks, while navigating the complexities of the hybrid cloud environment in the race to deliver a secure and quality experience to customers.