In short
Zarka Sultana, Principal Consultant at Sogeti UK (part of Capgemini), argues that a cloud migration only delivers better experiences for customers and employees when testing is designed around the way the migration is done. She matches specific tests to lift and shift, refactoring and replacement, recommends migrating in small agile batches, and walks through data, security and performance testing as the places where users feel the difference.
Key takeaways
- Moving to the cloud: understand the technology versus the testing approach.
- Choosing the right cloud testing approach can result in happier employees and longer customer relationships.
Why legacy systems push companies to the cloud
Sultana starts by setting out the options: private cloud behind a company firewall, public cloud hosted by a third party and shared over the internet, hybrid cloud that combines the two, and multi-cloud setups that large corporations use to spread workloads and stay available during outages. Whatever the choice, she says, essential testing must happen to avoid failure on deployment and poor user acceptance.
The motivation for moving is usually the cost and frustration of aging systems. In-house legacy software cannot keep up with newer technology, requires expensive maintenance of outdated hardware and software, and makes hosting mobile applications, APIs and file downloads difficult from a security and compliance standpoint. Performance is often the most visible problem. Sultana shares a moment from her consulting work: "I remember once I went to a client's site to do an independent piece of quality assurance for their product and the system took between 30 to 40 minutes to load every morning." (2:44) The result was predictable: "The staff comes into the office, turns on the computer, and then goes for a coffee break." (3:18)
Each migration path needs its own tests
Businesses reassess their requirements, map current architecture and dependencies to cloud providers, and choose a migration approach. Sultana argues that testing should be configured the same way, based on requirements, system functionality, business processes, integrated systems and the cloud solution itself.
For lift and shift, where an application and its data are rehosted on cloud virtual machines without changing code (typically infrastructure as a service), the risk of breaking code is lower, but complex dependencies and long outages remain a concern. Here the testing sequence covers data migration validation so nothing is lost between source and target, regression testing to confirm existing functionality still works, end-to-end process checks, security testing of user access and server vulnerabilities, and performance testing of load and runtime. Business users then run user acceptance testing against real-world scenarios.
Refactoring (often on platform as a service) and replacing with software as a service change more, so the testing changes too. "So I can see that during refactoring, rebuilding with the new code and replacing the database model, it is necessary to conduct a system test to confirm that each component of the new code operates correctly." (9:38) Because businesses rarely run a single application, system integration testing checks that components work together according to business requirements, and APIs are tested for performance, authenticated access and protection from malicious attacks.
Migrate in small batches and test as you go
Sultana makes a clear case against big bang moves. "So rather than a big bang migration, it is lower risk to adopt iterative and agile approach, that is to manage data migration to cloud incrementally in small batches." (10:56) That requires careful planning: categorising data, applications and integrated systems, then dividing the work into sprints based on workloads and dependencies, with each sprint focused on a specific element.
Standalone applications and databases with no dependencies can move first, with test execution running in parallel using a fail-first approach to gather feedback early. The same thinking applies to rebuilding code: "Similarly, during refactoring and rebuilding app, adopting agile approach for coding testing will prevent expensive time-consuming fixes later." (11:31) After migration, she recommends continuing with iterative Scrum cycles as applications are modernised.
Data and security: where users feel the difference
Poor data quality (inaccurate, inconsistent, duplicated, obsolete or lost records) leads to high operational costs, integration risk and interrupted business operations. Sultana recommends a pre-migration test rehearsal: risk analysis, source data assessment, workload type, recording the amount of data to verify later, timeframes for online and offline migration, schema checks on field names, types, lengths and mandatory fields, and data cleansing. Her simple flow runs from extracting data as it is, validating table row counts and cleansing, through transformation to loading into the target environment.
Security brings new challenges in the cloud, particularly the lack of standards in public cloud. "So cybersecurity is a shared responsibility between business and cloud provider to manage both risk and compliance." (14:29) Data should be encrypted during migration, disaster recovery backups should be in place, and regression tests should follow any rollback or restore. Access testing can include trying different password combinations to confirm an account locks after three attempts, alongside two-factor authentication checks. "So with that, we need to continuously scan server, endpoints, networks, and test application and databases." (16:17) She adds that employees need training on the cloud process and the new interface before migration.
Performance and the case for testing in the cloud
To keep the experience fast, Sultana lists load testing across low, normal and high traffic, stress testing to find the breaking point, latency checks and fault tolerance testing to confirm workloads stay available across geographical locations during disruption.
She closes by contrasting traditional testing, with its heavy spending on hardware and software to simulate users, with testing in the cloud. "Using the cloud to run tests can provide a compelling combination of lower cost, pay per use, and no upfront capital expenditures." (18:31) Add on-demand flexibility, better collaboration and faster time to market, and the end goal comes into view: "In summary, testing improves quality and performance of the software and eventually improves user experiences when they are very happy browsing faster without interruption." (19:15)
Watch the talk
"How Testing in Cloud Shifting improves the customer and employee experiences?" by Zarka Sultana, 20 minutes: https://www.womentech.net/video/how-testing-in-cloud-shifting-improves-customer-and-employee-experiences-zarka-sultana
About the speaker
Zarka Sultana is a Principal Consultant at Sogeti UK (part of Capgemini), a Digital Assurance and Quality Engineering organisation, where she provides expert testing solutions on client sites across telecom, energy, non-profit and government. She leads test management and quality engineering across the software delivery life cycle, bringing business, developers, testers and third-party stakeholders together to deliver continuous quality.
Questions this talk answers
- Why are companies moving from legacy systems to the cloud?
- What testing does a lift and shift migration need?
- What changes when applications are refactored or replaced?
- Why migrate in small batches rather than a big bang?
- Why run testing in the cloud itself?
Drafted with AI from the talk transcript. Editor review pending.