Transitioning to Agile/SCRUM: the impact on testing

This article has been published in Testing Circus – Volume 4 – Edition 12 – December 2013.

testingcricus

An increasing number of companies are using Agile/SCRUM to implement their software. It is quite a shift from traditional/waterfall development to the more flexible, less documented Agile/SCRUM approach. The transition often proves to be a challenge at some point.

In the book “Scaling Software Agility: Best practices for large Organisations” by Dean Leffingwell you can find some important processes that will change. Going through a couple of these processes, I will describe the impact of the transition on testing.

Changes caused by the transition

Measures of success

In traditional development the main measure of success for a project is usually on time delivery, but in Agile this changes to working code. This is a fundamental difference in measuring success. Instead of time driven, the project will be quality driven. Of course this has its impact on testing. When working code is the measure of success, we need to put more effort in providing working code. The only way for the team to find out that it is working is by testing.

In traditional development the focus for the different disciplines (such as business analysts, developers and testers) is on different aspects. In Agile the complete cross functional team needs to have its main focus on quality. To make this happen, it is important to spread the knowledge of testing across the whole team. This can be achieved by:

  • using pairing to pair testers with people in other roles to facilitate implicit knowledge sharing;
  • providing basic test training for the team members to explicitly focus on testing aspects of their roles.

Both of these methods can be executed by the tester in the team, but the latter of the two can also be executed by people outside the team. Testers need to communicate with all different roles in a project and help them to understand and apply testing in their context.

Management culture

Another important aspect that needs to change is the management culture. Where the keywords in traditional development are command and control, the culture needs to shift towards collaborative leadership and empowerment of teams. The impact on testing as a craft seems minimal, but the impact on the traditional test functions, such as test managers and testers, is large.

Test managers used to be responsible for test strategy, product risk analysis, test plans, test estimation, resourcing, etc. But how will this work in Agile/SCRUM?

  • Planning and estimation are a team responsibility.
  • Detailed product risk analysis upfront is not possible.
  • Teams need a degree of freedom, so extensive strategy and plans are uncalled for.

In short, the role of test management changes. The human resources aspect of management is more important. How to get the right tester in the right team and keep the testing knowledge of the testers up-to-date? This is done by knowing the testers and their needs. Test management needs to find ways to get the necessary information out of the different (SCRUM) teams in order to have a bird’s eye view on the testing process.

Keep in mind that a lot of the previous management responsibility shifts to the teams. This requires a high rate of trust in the people and keeping away from micro management. So management needs to let go some of the control and the people in the teams have more responsibilities and need to deal with this. Mind that not everyone will feel comfortable with this, so make sure to pick the right people for the different roles in the team. Next to that: not every team needs the same type of tester.

Requirements and design

The change in requirements and design is very big and testing needs to find a way to cope with this. Where we had big upfront design, we now have continuously changing, emergent, just in time documents. This impact on testing is felt at management level and at engineering
level.

In Agile, test management cannot identify the detailed risks, since there only is a high level, global set of requirements. To retain a risk based testing approach, we need different levels of product risk analysis, abstract up front and more detailed in the teams when more detail is known. So (test)management should be able to do a high level risk assessment at product backlog level, where the team will do detailed risk assessments at the sprint backlog level.

One of the main complaints by traditional trained testers in an Agile environment is the lack of upfront requirements and designs to use as a basis for their testing. Test cases need to emerge from discussions at the grooming or planning session. Testers need to start creating test cases based on these discussions before the requirements and designs are properly documented. By having testers and designers review each other’s products you have quality control early in the process. Test cases and designs prove to have a better match and any differences can be discussed with the product owner.

Note that a good product owner is indispensable for Agile projects. Only by good product ownership, the right product gets build.

Coding and implementation

Everybody is aware of the different phases in a traditional project. Testing takes place after coding, which in its place is after design. A good practice in Agile/SCRUM is the use of test driven development (TDD). Coding and testing then go hand in hand. This increases the quality and maintainability of the code. The shift to TDD is not always that easy. Developers often don’t like to write unit tests and did not receive proper training in how to do right TDD. When used incorrectly, TDD will take a lot of time and have no or little benefits. Testing can support once again by pairing to support the developers with applying white box testing techniques.

The short development cycles and the incremental approach require a lot of regression testing. Since regression testing takes place in every sprint, test automation will save a lot of time. The impact is that people with test automation skills are needed it the team and that we need to plan for the automation. If the testers are not able to automate the tests themselves, they at least need to know what should be automated and communicate this with the automation specialist.

Overall impact on testing

Basically, we still need to test. The craft of testing is still in place and we must not forget what we learned in the past. But we need to adjust and adapt to our new context: Agile. The quick and changing world of Agile development requires a more pragmatic approach to testing. No large upfront planning and documenting, but small pieces of functionality. Pieces that are manageable by the teams. Testing goes beyond the tester; it is part of the complete team.

Automation has become an essential part of testing to keep up with the development pace in Agile. This requires the testers to have more technical knowledge and better communication skills. An early start with automation, usually results in a better maintainable product, so
enough reason to automate!

Last but not least, it’s all about people! Investing in people and skills is needed to perform well in an Agile context. Provide training in testing for all team members and don’t forget to provide training in other disciplines for the testers as well.

How to sell software testing?

In the last two months I was testing software for Enrise via Polteq as a part of their development team The Impediments. Testing for them has been nice and instructive. The people at Enrise did not stop asking questions. Most of the questions I answered immediately, but some took a while to find the right answer. The question that lingered the most was: “How do we sell software testing to our customers?” I have to say that I’m very glad that Enrise asks this question! It shows me that Enrise sees the value and the importance of testing. The high quality software that Enrise delivers, can only be delivered when there is time set aside for software testing.

So how do you sell software testing?testing

  • Sell a project (testing is part of the project).
  • Only provide guarantee for production incidents when the customer pays for testing.
  • Make the customer aware why you value testing.

Basically, you do not sell testing, you sell quality!

Software development is practiced in teams. Testing is a set of tasks that need to be executed in the teams, so there is implicit time for testing. You need to see that testing is essential to quality software and the time needed to test has to be allocated. The software may appear more expensive to your customer, but this will pay off! In the long-term, you will see less disturbances of production when the software is professionally tested. Investing in quality up front is worth it.

You’re probably thinking: “How will testing add to the quality?” Well, in many ways… I’ll give you three:

  1. If you plan to test, you will need testable requirements. Ask questions on user stories until you have clear what your customer actually wants to have. Developers usually make more assumptions and think about the solution, where testers think about value for the customer. More questions in the beginning will result in an easier development process and a smaller chance on defects.
  2. A tester tests software instead of checking if the software works. A customer will use the software to see if it works, where the tester tries to find the edge cases and out of the box situations. A negative age is an unlikely situation, but a typo in a birthdate is quite plausible… The tester thinks of these situations, so when these typos arise in production the situation has already been covered.
  3. Testing is about mitigating risks. This implies that the risks need to be identified. Testing will create a better view on risks. So even risks that are not mitigated, can be communicated to the customer. Then it’s up to the customer to decide what to do: Invest more to mitigate these risks, or live with the risks?

In short: if you want quality products, you really need to test!

*This blog is also posted at Enrise

Experiences at EuroSTAR 2012 (part 3)

So now my experiences at Thursday, the final day of EuroSTAR 2012. See my other posts for the Tuesday and Wednesday experiences.

What Agile Teams Can Learn From World of Warcraft – Alexandra Schladebeck

As I am a World of Warcraft (WoW) player myself and a great fan of Agile, the title alone was enough for me to decide to attend this presentation. Alexandra has done a great job in pointing out the parallels between WoW and Agile, not only the benefits, but also the pitfalls. WoW is a massive multiplayer online role-playing game. As in all role-playing games, we see different races, classes and professions for our characters. Each combination will have its own set of skills, when characters form groups to be able to complete dungeons, they need characters with different skills on board.  Sounds familiar if you think of multidisciplinary teams right?  A team of individuals working together to achieve a common goal… When we go one step bigger and we set our goal even higher, we can do raids in WoW. When we try to do such a project, we need several teams that work together.

Wow & Agile

When these WoW teams start their quests, they need to do some planning. In this process the teams need to estimate what the harder parts will be and who will be responsible for which tasks. The proper equipment for the specific quest needs to be put in place and they all need to work together. For the communication most groups use a tool named teamspeak.  However in some points WoW is easier, since we can use dragons for fast transportation and portals to get all the people easily at the same place.

The slide on what to learn was really interesting and therefore I added it to this post (click on it to see a larger version). Additional it is important to learn to do more that just your specialization. Just keep working in teams fun, this is applicable for both WoW and Agile teams. And finally learn to rely on your team, since you can’t kill the boss on your own 😉

Testing the API Behind a Mobile App – Marc van ‘t Veer

Polteq was happy enough to have my colleague Marc also selected with his presentation on testing an API. Marc used all his experience at T-Mobile to guide us through testing an API. He started off by explaining why T-Mobile wanted an API behind the mobile Apps. Since T-Mobile has a place where you as a customer can log in and see your calling and texting bundles. A lot of  independent App creators created App that allowed T-Mobile users to do this via their mobile phones by using screen scrapers to get the information to display. Whenever the App malfunctions – broken or incorrect data – the users blame T-Mobile for this. Even worse, the App creators also point a finger towards T-Mobile. So T-Mobile decided to decouple the content and make App creators use the API to get the content. This allowed T-Mobile to be more in control of the data and the meaning of the data.

So how to test an API? Marc starts off by showing us some risks involved with API’s:

  • It’s impossible to know up front how the API will integrate with the external Apps
  • There is a big variation in the data that will be provided by the API
  • There is no full control on the end-to-end process
  • The API may be used incorrectly

To be able to do early integration testing, T-Mobile used a prototype App and used dogfooding during development and system test. An adapter was created to let the API communicate with the back-end, so integration with T-Mobile’s back-end could be tested. This adapter also served the My T-Mobile pages, so the data on these pages could serve as an oracle for the data in the App. In testing they noticed that caching was not properly working. Since at first a single security key was used for all users. So when testing an API, make sure that you test with different users that have different authorizations. Another defect that showed, was that the HTTP-Statuses were not informative enough for the App. The API then was edited to supply extra information, so the application could provide the right information to its users. The T-Mobile data provided some difficulties of itself, since there are multiple types of bundles and each bundles has a maximum number of units that can be used. However the tag was used for different entities. One time it meant minutes, the other time it was a number that showed the number of text messages you had left or a combination of these two.

To test the API, the testers needed a lot more technical skills, since testing involved a lot of command line functionality. To actually test the API properly, automated regression testing in production was needed. Do not forget to apply the testing techniques that have proven to be valuable over the years in this new context.

In the end a good API was introduced, but people still see T-Mobile as responsible when an App malfunctions.

The testlab – Bart Knaack

The testlab cannot be absent in my experiences. How great is it to actually do some testing at a testing conference! In addition to the website and application testing, this year we got to play with Lego Mindstorms 😀 The first task was to find out what the provided car would do. It used a light sensor to read different colors and when it read the color it would do an action based on the color. After determining the actions that relate to the colors it was our tasks to see if these would hold. Of course there were bugs present! I don’t want to spoil the fun for future use of the Mindstorms for testlabs, so I won’t mention the bugs here. As you can see on the image, I earned the “I logged my bug in the testlab”-button. As simple as the reward seems, I made me happy and delivered a smile when I received it.

Testlab buttons

Experiences at EuroSTAR 2012 (part 2)

In the previous post I described my experiences at the Tuesday of EuroSTAR 2012. In this post I will continue my EuroSTAR experience with the Wednesday.

Changing Management Thinking – John Seddon

A nice quote from this talk about changing management thinking: “The primary cause of failing is management”. Managers tend to make decisions that are not in the benefit of projects. For example when you want to decrease the costs, managers start managing on costs… This actually increases the costs in most cases. However, when you manage on value, this will more often decrease the costs. A very useful story that John told us, was about chicken wings and spare ribs. Management of a large chain of restaurants decided to replace the spare ribs as a starter with chicken wings, since the chicken wing had a larger margin. Customers were disappointed and asked the waiters if they could get a small portion of ribs (still available as a main) as a starter. The waiters want to please their customers, so they say that it is possible. Then the fun part will start… The waiter needs to put the starter in the cash register and since there is no starter of spare ribs listed, they choose to file it under chicken wings. Since management reads the registers and sees that chicken wings are sold very often, they order more chicken wings for their restaurants. A fine example of failing management.

Adventures in Test Automation – Breaking the Boundaries of Regression Testing –

 John Fodeh

John provided information on automated monkey testing. The presentation was supported by using some scenes from the IT Crowd to inform us on automation.  Automated monkey testing p roved to be an easy to understand concept: by randomizing each step, you are simulating monkey testing. The problem of course is that it is easy to miss out on obvious defects, it does not effectively emulate real scenarios and debugging lng test runs can be quite a pain. They felt the need to create more intelligent monkeys by creating somewhat more expectable behavior via the use of state tables with probabilities per action.

Evolving Agile Testing – Fran O’Hara

After a short introduction on Agile and SCRUM, Fran started off on requirements. When we start to talk about user stories, we should try to find out about acceptance criteria for the story. This serves several goals:

  • Define the boundaries for a user story/feature
  • Help the product owner to find out what it is that delivers value
  • Help the team gain understanding of the story
  • Help developers and testers to derive tests
  • Help developers when to stop adding functionality to a story

Fran reminds us to keep these acceptance criteria at a relatively high level, so do not lose yourself in too much details. Detailing will be done in e.g wireframes, mockups or validation rules. Another place where we find detailing is in the automated acceptance tests. Try to find examples that support your acceptance criteria.

Next Fran stresses the fact that we still need test strategy in Agile. We need to think about the minimal tests in the sprints (automated unit, automated acceptance, manual exploratory) a

nd sometimes need to do some additional testing e.g for non-functionals, feature integration or business processes. The testers themselves need to have broad knowledge (more than just testing) and deep knowledge in testing. This requires a ‘technical awareness’.

Testing of Cloud Services; The Approach: From Risks to Test Measures – Kees Blokland & Jeroen Mengerink

Kees and myself presented Cloutest® our approach to testing cloud services. We started off with an introduction to cloud computing to set the context. To properly introduce the concept, we decided to use the definition provided by NIST. After this into our approach. We identified 143 risks that arise when using cloud computing and grouped these risks into categories:Cloutest-Eurostar

  • Performance
  • Security
  • Availability & Continuity
  • Functionality
  • Manageability
  • Legislation & Regulations
  • Suppliers & Outsourcing

Since 143 risks is quite a lot, we decided to give a limited set of examples of risks and detail these. For instance there occurs a performance risk, since a cloud service usually has several customers. So it’s not only you as a customer that is putting load on the service, but also other users of other customers. This will influence the performance of the service. Imagine your webshop hosted at the same hosting provider that hosts WikiLeaks… The huge amounts of traffic that a new publication on WikiLeaks will generate, might result in your webshop not being available due to performance problems of the service.

With testing we provide methods to mitigate risks, so that is what we did too. The good news is that we can still use a lot of what we have learned over the years. Some techniques need to be tweaked to fit in the cloud context, but they are very useful. Next to the tweaked measures, we also describe some new measures that we have used at our clients. We grouped the measures too:

  • Selection
  • Performance
  • Security
  • Manageability
  • Availability & Continuity
  • Functional
  • Migration
  • Legislation & Regulations
  • Production

How to test the scalability of a cloud service??? Providers promise scalable services and customers pay per use, so if you need more, you will get more. With traditional load testing, we can gradually increase the load and see how the system responds. This can be applied to the service too, but it will scale. You will see the point where the scaling starts in your response times, they will drop when more performance is added at the service. Check around the boundary of the scaling point to see if the billing is also scaled.

We see that test starts earlier, the scope is wider and testing will not stop in production.

Inspirational Talk: Sky is not the limit: Copenhagen Suborbitals – Peter Madsen

The inspirational talk was very nice, however not very test related. It showed us that with the right vision and perseverance you can reach goals that seem to be unreachable. Peter showed us how he built a homemade submarine and a homemade rocket.

Noordertest 2012

NNOTLast week the Noordertest conference took place in Groningen. With three groups of five parallel presentations and one keynote, each of the 160 attendees could find something interesting.

Polteq provided two presentations for the conference. I presented about Test Improvement for Agile together with Edze Knol and Ruud Teunissen presented about how to properly do test automation. Edze and I were in the first group of presentations, so after a general introduction to the conference we had to kick-off.

Test Improvement for Agile proved to be a hit, since we had more people than we had seats 🙂 We started with a short introduction to Agile and SCRUM, followed by a short introduction on test improvement. After the introduction we provided some more depth information on three of our key areas:

  • Teamwork
  • Test management
  • Defect management

In teamwork we seek for collaboration, trust and the willingness to work outside your comfort zone. About test management you can read more in my previous post and defect management should be to support the team in stead of the business.

After our presentation I attended “The fragility of agility” by Lloyd Roden. It was nice to see that he pointed out the same groups of improvable items that we dealt with in our presentation.

Parallel was the presentation about proper test automation by Ruud Teunissen. He told the audience that we need to make sure that test automation bridges the gap between testware and the system under test. Make decisions on what you want to achieve and not on e.g. the tools that are already present in an organization. Remember that test automation is a form of development and should be treated as such.

Next was a workshop on how to ask questions. Main points were to make sure that you ask the question you really want to ask and then listen to the answer. Don’t add your own information while asking questions, so you will get the real answer in stead of what you want to hear.

Finally the keynote – also by Lloyd Roden – about challenges in software testing. Here Lloyd presented eight challenges in the software testing world. Learning, skills and communication where of course part of the challenges.

I really enjoyed the conference!