LogiGear

14Devices That ThinkEmbedded system testing is very different than applicationtesting. With the explosion in things that think and ubiquitouscomputing/sensors and devices that interact, andoften provide some components of big data, new test methodsare needed.As an overview of some of the complications: How to generate a function call and parameters: Need for new tools, APIs, various input systems. Need for virtual environments. Separation between the application developmentand execution platforms (the software will not bedeveloped on the systems it will eventually run on.This is very different from traditional application development). How to simulate real-time conditions.Difficulty with test validation points or observed behaviors: Parameters returned by functions or received messages. Value of global variables. Ordering and timing of information. Potentially no access to the hardware until finalphases of development. All these will potentially complicate test automation.With the added worry that many embedded systems aresafety-critical systems we will need technical training, moretools, new thinking, new methods and more sophisticatedtest automation for devices that think with no UI.Big DataBig Data methods are not jump running a few SQL queries.Saying big data is difficult to work with using relational databases,desktop statistics and visualization packages is anunderstatement.Instead, "massively parallel software running on tens, hundreds,or even thousands of servers" is required. What isconsidered "big data" varies depending on the capabilitiesof the organization managing the dataset, and on the capabilitiesof the applications that are traditionally used to processand analyze the data set in its domain. For some organizations,facing hundreds of gigabytes of data for thefirst time may trigger a need to reconsider data managementoptions. For others, it may take tens or hundreds ofterabytes before data size becomes a significant consideration(www.en.wikipedia.org/wiki/Big_data).Correlating global purchases at Wal-Mart with direct advertising,marketing, sales discounts, new product placement,inventory and thousands of other analyses that can bepulled from the 2.5 petabytes of data creates serious testingchallenges. Testing the data, the algorithms, analysis,the presentation, the interface, all at the petabyte levelsrequires software testing by people like you and I.This scale of testing will require huge virtualization efforts,a new test environment set-up, management skills, datamanagement, selection and expected result prediction, aswell as the obvious data analysis skills. There will be heavytool use in big data testing. Quick facility in aggressive automationis a pre-requisite.MobilityMuch has already been written about the increased needsfor mobile testing. I will focus on two less discussed ideas.There is less and less focus on the phone aspect of mobiledevices, and more focus on intersecting technologies. Fromcommon intersections such as Wi-Fi, Bluetooth and GPS tonewer and high growth intersections of RFID, the currentlyhot NFC (near-field communication), ANT+ (used mainly formedical devices), QR Code (quick response code) and agrowing variety of scanners. As with things that think, learningnew technology, new input tools and methods, capturingoutput for validation and observation and complexitiesfor test automation are demanded with fast ramp-up.Security and performance testing are two commonly neglectedaspects of mobile testing-. Much lip-service is givento the need for these test methods which need differentplanning, skill, expensive tools and time to execute. In today’sfast-at-all-cost delivery environments these test methodsvery often get postponed until a crisis occurs (checkout ―Mobile Testing By the Numbers‖ for information onhow much mobile apps are performance tested). Testersneed to be excellent at risk assessment and planning forthe times when these are added back into the test strategy.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