LogiGear

More documents

Recommendations

Info

12and early 2000’s has totally reversed. Now the focus is onleaner processes with the best metric of team success beingthe delivery of working software. The only metric in Scrum isburndown, velocity is secondary, and bug counts may bemoot and turned into user stories. This may be extreme foryour organization, but it is definitely the movement.Joint Ownership - Ownership of QualityIt has been a long time since people who test gladly got ridof the QA (quality assurance) job title. However, some unawarepeople continued to believe in those who test guaranteedquality rather than measure it, report risk, find bugs, orany other achievable tasks. The Agile movement has finallydone away with the illusion with more Teams today espousingjoint ownership of quality. There are more upstreamquality practices: code review, unit testing, CI, better userstory and requirements from need for customer satisfactionand closer communication with test teams. Following theScrum framework, most teams now have the Product Owneras the ultimate owner of quality.SpeedThe speed of development today is a changing process.The pressure to deploy rapidly and simultaneously on multipledevices or platforms— the ease of distribution, marketpressure, cost pressure, do more with less pressure— ispushing some teams to abandon or reduce even some universallyaccepted practices, such as hardening or releasesprints, to accommodate the need for speed. Abandoningprocess for speed is as old as software development itself,but the pressure is greater today. This is risky when testinglarge and complex systems where test teams need highvolumetest automation and excellent risk reporting skills.SaaS/CloudSoftware development has been greatly impacted by theuse of ALMs (application lifecycle management tools). Thisis not the place to detail ALMs, (I went into great detail onthe subject in a past article) except to say that their usetoday is bigger than ever and growing. That these tools aremore integrated with testing, automation and test managementis new.There are evolving process changes in the Cloud, particularlyin SaaS. With the immediacy of deployment, rapid change,and ever-smaller or non-existent architecture, some Agileteams cannot keep up. With the speed SaaS products aremade available to all customers, patches, hot-fixes or rollbacksare particularly problematic. As always, large-scale testautomation is a team’s most important defense.ALMs can greatly facilitate communication with offshoreteams. Continuous Integration (CI) is a pervasive XP practicetoday. The best ALM tool suites have complete CI tooling(source control, unit test execution, smoke test/UI/automated build validation test execution, autobuild, traceability,and reporting) or easy integration for these.Test ProcessTesting often lags behind development in most organizationsand test teams are still responding to Agile. Many testteams are struggling in ScrumBut processes. Most tests aretrying to prevent ―handoffs‖ and waste rework as well asbeing flexible and responsive. As noted above, test teamsare still figuring out test practices on Scrum teams and inXP practices. This has been the subject of many articles at<strong>LogiGear</strong> Magazine.The biggest test process impacts, or game changers totesting in Agile/XP, are CI and the need for high volume testautomation. These are detailed in the following ―Practices:Methods and Tools” section.Practice - Methods and ToolsMethodsThis area is leading the most dramatic changes in softwaretesting.Massive Automation– Smoke Tests and Regression TestsThere are two foundation practices that have had a profoundimpact on the testing landscape. Continuous Integration—mistaken sometimes to mean only auto build—is theautomatic re-running of unit tests, the test team’s smoke orUI automated tests, and reporting changed code, fixed bugsand automation results. This process is significant to thenew Agile speed of development, requiring rapid release,and is dependent on test teams having significant smoke orbuild validation tests as well as traceability. Using an ALMsuch as TFS (Team Foundation Server), for example, thathas all the pieces for CI built-in, the reporting can detail thecode changes and the bug fixes. Gone are the days of gettingbad builds, or good builds with no idea what changed.The test team must have a great automated smoke test tohave a true CI practice.W W W . L O G I G E A R M A G A Z I N E . C O M F E B R U A R Y 2 0 1 3 ǀ V O L V I I ǀ I S S U E 1
13Regardless of what other practices you do, or advancesyour team is making to keep up with the new paradigms insoftware development, you cannot succeed without massiveautomation. As we know, old, high-maintenance, random,automation methods will not work today. Automatedregression tests are not only for speed, they are necessaryfor confidence, coverage, focus, information and to free usup to do more exploratory discovery. This last point is oftenlost in the new Agile development paradigms. With reduceddesign time and vague user stories, test teams need todiscover behaviors (intended or unintended), examine alternativepaths and explore new functionality. No one has thetime to do this without massive automation.We also know that scripted, old style automation does not work.I suggest you read Hans Buwalda’s article on Action basedTesting for the most advanced methods for test automation.The Cloud and SaaSThe CloudThe biggest impact of SaaS on test methods is the necessityand growing importance of performance, load, security,and scalability testing, in addition to having our hands fullwith functionality testing. Where load, performance andscalability testing have often been postponed due to themethod’s reputation of being expensive, slow, nice-to-have,optional, etc., there needs to be crucial discussion beforecompromising security testing.The value proposition behind SaaS is: let ―us‖ manage theservers, databases, updates, patches, installs and configuring.Everything your IT team used to do when you boughtsoftware, you can now buy as a service. Implicit in this proposalis - ―We‖ will manage and secure your data. The biggesthurdle most companies face in letting someone outsidethe company manage and maintain corporate intellectualproperty is data security. Security testing is not optional.In some cases, security testing is performed by specializedteams with their own tools. In most cases I see, thetest teams are integral to security testing in:Test case design.Data design.Stress points.Knowledge of common and extreme userscenarios.Reusing existing automation for security testing.VirtualizationThe primary impact of virtualization is the ability to dothings faster.Virtualization makes managing test environments significantlyeasier. Setting up the environments requires thesame amount of work, but regular re-use becomes significantlyeasier. Management and control of environmentsthrough virtualization is now a basic tester skill for:The common use of the cloud in software developmenttoday is virtualization of environments to quickly sharetools, resources and run simulations of both hardware andsoftware. This becomes more relevant in the context of thecurrent explosion in the mobile applications market forsmartphones and tablets. These applications require shorterdevelopment cycles and automated virtual regressionsmaking cloud testing cost-effective and efficient.SaaSSoftware-as-a-Service presents unique test method challengesthat are different from process changes with SaaS.A positive aspect of SaaS testing is the ability for teams tocontrol the production environment. There is no server sidecompatibility testing and even client-side access can berestricted and controlled.Executing tests on multiple virtual servers, clients,and configurations using multiple defined data setsincreases test coverage significantly.Reproducing user environments and bugs is easier.Easy tracking, snapshots and reproducibility to aidteam communication.Whatever resources are constrained, such as database,service or tools, your test automation andmanual execution can be virtualized for independent,faster, shared or not, independent execution.Facilitate faster regression testing economicallyusing many virtual machines.Virtualization frees-up your desktop.Performance bottlenecks and problems are common withvirtualization test projects. Automation may be more complexto set-up. Daily deployment of new builds to multiple (think100s) of VMs needs skill to run smoothly. To be effective,your team will probably need a virtualization or tool expert.W W W . L O G I G E A R M A G A Z I N E . C O M F E B R U A R Y 2 0 1 3 ǀ V O L V I I ǀ I S S U E 1
Page 1 and 2: SHOWCASING THOUGHT LEADERSHIP AND A
Page 3 and 4: 3I n thi s Issue4 I N T H E N E W S
Page 5 and 6: 5“Blogger of the MonthWhat Could
Page 7 and 8: C over St or y7The Changing Landsca
Page 9 and 10: 9Virtualization can be viewed as pa
Page 11: 11People - Skill, distributed/offsh
Page 15 and 16: 15New Input TechnologiesIn the past
Page 17 and 18: 17A thorough understanding of netwo
Page 19 and 20: Glossar y : Sof tware Tes ting Land
Page 21 and 22: 21In addition to the main body of t
Page 23 and 24: 23W W W . L O G I G E A R M A G A Z

LogiGear

Create successful ePaper yourself

Delete template?

Save as template?