Test Automation Best Practices (with SOA test approach)

Software Test Automation-Best Practices Leonard Fingerman Email: lfingerman@gmail.com

Why Automate Testing ? Improve software quality while reducing time to market … not simply increase test coverage Improve reliability and consistency of testing processes 3. Allow manual testers do more complex & more suited for human kinds of testing (exploratory, usability etc.) Enforce SLAs (service level agreements) Bottom line is save your company $$$

Common Types of Automated Testing Unit Testing (xUnit, TestNG …) Performance Testing (LoadRunner, JMeter …) Web Service Testing (SOAPUI, HP ST, iTKO LISA…) Security Testing (WebInspect, WireShark…) Functional Testing (via UI, API, Web Services… ) Component- level System- level Integration- level

Common Functional Automated Test Tools No GUI Object repository Only web-based application support Tool support is limited Good browser support Good language support (Java, Ruby,C# ) Can be easily extended as JUnit suite Open-source (no license cost) Selenium RC & IDE VisualBasic scripting is limited No IDE (changing in release 10) License cost Supports Web 2.0, Java or .NET applications Full GUI Object Map repository Seamless integration with QualityCenter HP/Mercury Quick Test Pro (QTP) Insufficient browser support License cost Built as Eclipse Plug-In with full IDE, Java support & Source Mgmt Supports Web 2.0, Java or .NET applications Full GUI Object Map repository IBM/Rational Functional Tester (RFT) Cons Pros Tool

Typical Test Automation Pitfalls You selected a Tool…now what? A tool is supposed to do it all, right… ? Don’t fall into tool vendor’s sales pitch …remember Capture & Playback is not real test automation According to recent IDT study ( www.idtus.com )

Tips for Test Automation Strategy * Essentially automated test system is a piece of software designed to test your application  treat it as such Evaluate an automated tool POC (proof of concept) based on your native apps Choose a tool that fits your business requirements & internal technical expertise Plan Establish goals & objectives ; budget; use or hire competent resources Be ready to have quality metrics to validate the ROI (return of investment) Design Keep in mind the user community of TA system (Usability) Choose an architecture that reduces maintenance & extensible Implement Practice OOD principles of reusability & modularity (employ encapsulation & abstraction) Introduce version control of your code Establish code review process Test & Maintain Unit test your code Constantly re-evaluate reasons of maintenance & improve TA architecture to minimize maintenance. Application testability usually plays a big factor. Evaluate defect rate with root cause analysis

Note: the testing tool is abstracted by higher-level framework component library

2 nd Generation Automation Frameworks “Data-driven” Automation is data-centric Test Tool – dependent (due to flow control in the script) User defines just data sets to drive tests with Data is de-coupled from script & defined in external data source (Data pools, Excel spreadsheets etc. for data sets) Flow control (navigation) is normally done by the test script not via the data source Example: data set exercises creation of new sales accounts functionality; stored in a DB table account_data MA Boston 02111 2332 Washington st Mick Jagger Jr. RollingStones Inc. GA Atlanta 30075 5775 Main st Phil Collins Genesis Inc. State City Zip Street PrimarySalesPerson CompanyName

Java Code Snipet of Data-Driven RFT Script public void testMain(Object[] args) { String name;String primarySalesPerson;String address;String zip;String city;String state; String query = "Select * from account_data";String AccessDBPath = Constants.FL_DB_PATH; //Data abstraction class DataHandler – methods process data and return data values DataHandler dlrAcctData = new DataHandler (AccessDBPath,query); ArrayList listDlrAcctData = dlrAcctData.getRowsOfData(); try { for ( int i=0;i<listDlrAcctData.size();i++){ coName = dlrAcctData.getValueForColumnName(i,“CompanyName"); primarySalesPerson = dlrAcctData.getValueForColumnName(i,"primarySalesPerson"); Type.text(CustomerCreate. field_name (),coName ); Click.object(CustomerCreate. image_primarySalesPersonIcon () ); Type.text(text_employeeModalPanelCommonN(),primarySalesPerson); … } dlrAcctData .closeDB(); } …

3 rd Generation Automation Framework “Keyword-Driven” Automation is action-centric Tool-agnostic (abstracted from tool if implemented correctly) Users de-compose test cases into granular & re-usable Action Keywords The idea is for non-coders to be able to author automated test cases with Action Keywords User defines flow control of the test via Action Keywords Example: Test Case “Verify Checking Account Balance” Enter Username and Password and Click submit button (entire step 1 may become action Login) Enter “Phil Collins” as a Sales Person and Click Submit button Verify the Sales Person was successfully created and Logout So you may want to choose the following re-usable action keywords: Login ; Enter; Click; Verify; Logout

Benefits of “Keyword-Driven” Approach Test Case definition is abstracted from the tool and can reside in external data sources (DB tables, Excel spreadsheets ...) or Test Case Management Tool that supports keyword based test cases (QualityManager, QualityCenter …) Lands itself well with Agile methodology Keywords can be extended to emulate domain language Customers and non-technical testers can define tests that map to business workflow more easily Example of Keyword-driven test case: .url:=.*createdSalesPerStatus.html .id:=SubmitSalesPer Button Click CreateSalesPerson Click Submit 3 UserLogin Module .value:=User Created Successfully .id:=CreationStatus DIV VerifyExists CreateSalesPersonStatus Verify Sales Person Creation Successful 4 .value:=Phil Collins .text:=SalesPerName Field EnterText CreateSalesPerson Enter New Sales Person data 2 .text:=Login Successful .id:=LoginSubmit N/A Login Home Login 1 Expected Object Type Action Page Description Step

Java Code Snipet of Keyword Library Class package framework.keywords; … public class Verify { … public static boolean exists(GuiTestObject object) { logInfo("Action: VerifyExists"); try { object.exists(); logInfo("Object Found"); passed(); result = true; } catch(NullPointerException e) { logError("Object Not Found"); failed(); } return result; } }

Hybrid Test Automation Framework Combines the best of both worlds (Keyword-driven + Data-driven) User defines data sets to drive tests with User also defines flow control of the test via action keywords Data is separated from script and is stored external data source (DB tables, Excel spreadsheets, XML for data sets) with action keywords in addition to generic and test case specific data sets

Framework Type Recommendations Takes more effort to design & implement and may require more coding expertise Data driven iterations may not be implemented Non-coders can author automated test cases with action keywords Makes it possible to scale Test Automation Creation & Execution User defines flow control of the test via action keywords (Login, Click …) Creates Domain Language Powerful technique for Agile Testing Keyword-driven Takes the most effort to evolve and thus requires the most investment with design & implementation Non-coders can author automated test cases with action keywords User defines flow control of the test via action keywords (Login, Click …) Users can data-drive the iterations of test execution Data is separated from the script Hybrid Flow control is not abstracted & implemented in the script Difficult to use for non-coders Takes less time to design & implement compare to keyword or hybrid (less of initial investment) Data-driven Cons Pros Framework Type

Test Execution Phase Based on personal experience 20 step UI-driven test case would take about 2 minutes to execute with 1 run box To scale total test execution time linearly add run boxes to the pool, e.g. with 5 run boxes of the same HW grade you can execute up to 100 steps in 2 minutes Now compare how long it would take a manual tester to execute 100 steps  that alone will self- justify value of test automation

Service Oriented Architecture Definition: SOA is an architectural style whose goal is to achieve loose coupling among components which leads to greater re-use of business logic As a result , SOA Streamlines applications Standardizes multiple system interfaces between consumer and backend systems Centralizes commonly used functions Provides a communication layer between the application and backend systems After SOA Application A DB Application B DB Application A Presentation Layer Application A DB Before SOA (3-tier architecture) Application A Biz Logic Layer Application B DB Application B Biz Logic Layer Application A SOA Services Application B

Web Service versus GUI-based Testing Web Service has no preventative input validation like the GUI Greater reliance on Data – data handling under positive, boundary, negative, null conditions Impact: Single low-quality service can impact multiple consuming clients Consuming Client GetWeatherInfoByLocation (‘Atlanta’,‘MA’,’Mexico’) Web Service ???

Challenges Testing SOA Testing Middleware/SOA is inherently more complex than testing traditional enterprise GUI-driven applications : Middleware is head-less (no GUI) and thus testing is similar to API Middleware is not built with viewer in mind hence messages require and return far more data (like XML) than human-oriented user interfaces That means: Greater need for SOA architectural and developer skills throughout the lifecycle of testing in a SOA Tester (WS-I, WSDL, SOAP, XML, Java, .NET) Greater need for collaborative cross-functional test teams The good news is: SOA testing will run it’s own Full Life Cycle Testing independent of the application testing Continuous testing across the delivery lifecycle must be part of quality management strategy – automation plays greater role especially with regression and performance

Layered Testing Approach Like an onion a deployed web service has many layers How do we approach testing complexity of web service layers ? A layered testing approach is also required (core component-level to integration-level to system-level testing). More rigorous testing with completion of proper testing at each level Greater reliance on automated testing

Automating Testing of Web Service Components HP Service Test (WSDL-Generated TCs) Inbound Request Service Operation under test Ex: NameandAddress Service 1 Using SOA test tool (HP Service Test) generate inbound service request based on SUT (Service Under Test) artifacts i.e. WSDL, XSD schema, design documents and familiarity of source code. Drive inbound requests parameters with negative, positive, boundary and null data values. Configure outbound response of SUT to point to calling client ( HP Service test) Response is validated via checkpoints in SOA Test Tool (HP Service Test) addressValidation 1 2 Outbound Response 2 Emulated Service Subscriber Or Calling Client Web Service Operation under test 3 3 Test/Defect Management System HP Quality Center Test Exec Result Log Validate

Methodology to Reduce Test LifeCycle and Accelerate Parallel Development - iTKO VSE DELAY Web Services (under construction) Time Time Front-End Dev & SOA Test Teams DELAY Back-end under development/ 3 rd Party Service dependencies Virtualized Web Services based on WSDL Web Services DEV & TEST Teams Web Services DEV Team Front-End Dev & Web Service Test Teams VSE Before After VSE

Calculating the ROI Time Cumulated Costs ($) Initial Investment Automated Testing Manual Testing Cumulative Cost Saving ROI= Net Value / Investment Cost * The maintenance cost of TA is usually not trivial but if you keep enhancing your TA framework, improve application testability and overall testing processes high ROI is very much achievable

STAF – Software Test Automation Framework http://staf.sourceforge.net/ Automated Testing Institute http://www.automatedtestinginstitute.com/home/ LinkedIn User Groups http://www.linkedin.com/groups?home=&gid=159501&trk=anet_ug_hm http://www.linkedin.com/groups?home=& gid =46748&trk= anet_ug_hm Resources

Q & A Leonard Fingerman [email_address]

Bibliography Implementing Automated Software Testing: How to Save Time and Lower Costs While Raising Quality by Thom Garrett, Elfriede Dustin; ISBN-10: 0321580516 ; ISBN-13: 978-0321580511 Agile Testing: A Practical Guide for Testers and Agile Teams by Lisa Crispin, Janet Gregory ISBN-10: 0321534468 ; ISBN-13: 978-0321534460 http://www.stpcollaborative.com/magazine

Test Automation Best Practices (with SOA test approach)

Recommended

More Related Content

What's hot (20)

Similar to Test Automation Best Practices (with SOA test approach) (20)

Recently uploaded (20)

Test Automation Best Practices (with SOA test approach)