Big Data, Small Sprint

W16     Start Time : 13:45     End Time : 14:30

WATCH: Adam introduces his talk in a short video

For 7 years I have worked on the testing of a data storage and query system within the context of an agile development team. Initially targeting single server data archiving implementations, we have seen the data storage requirements of our customers increase exponentially to the trillions of records and petabytes that now constitute the “Big Data” requirements of multinational organisations.

In this presentation I’ll examine the massive growth of the customer data storage requirements and the demands that these have placed on us in testing. I’ll look at how an agile approach has presented unique challenges when attempting to test to massive scale within time-boxed sprint iterations. Without time or the budget to recreate the size of our largest customer implementations, we have had to use innovative approaches to testing to ensure that scalability and performance demands are met. By gaining a thorough understanding of the software we have been able to apply tactical scaling at the appropriate levels of the application to gain confidence in the performance of the wider system.

Through treating our own test automation as a software development activity, rather than a tool procurement, we have been able to incorporate new ideas over time, including parallelisation, iteration and randomisation, to keep pace with the constantly changing demands of our supported topologies. We have also been able to shorten lead times to allow us to quickly and reliably generate large scale test installations through some innovative approaches to automation setup.

I will examine the challenges of testing a technology that operates across both traditional storage technologies as well as the newer and more volatile big data technologies of Hadoop HDFS and MapReduce. I’ll review both successes and failures in tackling the big data problem as it has driven our testing ideas and how we are continuing to grow to keep pace with the ever increasing storage requirements of modern organisations.

  • Speaker

  • Adam Knight - , RainStor, United Kingdom

    Adam Knight has been testing data storage and analysis software for over 10 years, with 7 of those spent working in an agile team.

    Adam is an enthusiastic exponent of exploratory testing approaches backed by discerning use of automation. He is a great believer in creating multi-skilled teams based on rich and unique individual skill sets. At his current employer, RainStor, Adam has overseen the testing and technical support of a large scale “Big Data” storage system from its initial release through to successful adoption in some of the largest telecommunication and financial services companies in the world. He writes at www.a-sisyphean-task.com.