A Microservices Outcome: Testing Boomed

Key Takeaways

[Music] you're watching the new stack makers a podcast for people who develop deploy and manage at scale software for more information and articles about at scale Technologies please visit the new stack.io now enjoy the show we're a team of experienced technologists creating development and testing tools which help teams build Cloud native applications easier better faster cubeshop is backed by the most successful investors in the software industry who see kubernetes as the foundation for building the software of Tomorrow Insight Partners is an investor in Cube shop and the new stack hey everyone today we're talking about testing Cloud native testing and it's a topic that I'm really interested in exploring today with Bruno Lopez whose product lead at Cube shop hi Bruto how are you nice to meet you and thanks for having me I'm doing fine and you I'm awesome I'm awesome you know software testing is changing so fast everything's changing so fast as we'll we well know you know the complexity that comes with Cloud native Technologies is leading to new approaches for testings and and it's affecting the personas the the people who are actually developing their Technologies be they developers administrators whoever it might be so to me it feels like a little bit of like a Renaissance there's a Renaissance in testing and I'm curious to understand that story what is that story about testing in kubernetes and its Origins and how you see it leading to new creativity and new ways that Engineers are deploying their Technologies definitely so it's very true that cloud is a very fast uh moving world and all the Technologies are always appearing all the time and all developers from all areas like from development infrastructure operations testing everybody is impacted right and what is these kubernetes that we talk about it's uh the way people deploy applications to the cloud also has been changing a lot especially in the last let's say five to ten years and a lot of companies are adapting to these Cloud Technologies but not on all aspects for example some people use clouds not to deploy the applications and ask them in kubernetes so kubernetes is the place where people host uh duplications to serve to their own customers so they adapt that part which is deploying the applications but the testing aspects of it sometimes just follows after so it's it lacks it lags a bit behind so what it means is that the testing suffers and sometimes when when companies have priorities when they just want to push shots like new features to the product and just want to move fast and the methodologies and the process they have inside the company didn't move as quickly sometimes X is a bit neglected and so there's like the what we are seeing and so kubernetes is like an amazing framework it's amazing platform allows companies to do a lot of things but you know people need to adopt uh from everybody in the organization that is involved with the development of new software also needs to adapt as well so when did it get started like what do you trace it back to you say five to ten years ago you know kubernetes emerge in the uh you know 2015 or so 2016. and it changed a lot of how we think about Enterprise architectures is that the origination do you think for how we think about testing kubernetes what are some of the things that you reflect upon it's the way people um think about developing also their applications so kubernetes as an orientation to something called microservices so before that people built huge applications with a lot of process to test them testing was very complicated took a lot of time and then some very intelligent people decided Well what if we break these applications into small pieces that work very well together and they are easy to scale they are easy to test they are easy to developer to develop and then they build like you know these architectures to to sustain those things and testing also improves a lot because you can test small pieces of software in an easier way okay and so the microservices story really emerges here what did it lead to what what has it led to for for engineers make made more people involved in testing when when you have a big application that is very hard setup for example if you want to run tests you have to have a dedicated system for it right it's cannot run on your computer if let's say you have a big organization and for for your organization to run their product it's a very very big thing um then for you to run it it's like complex you cannot just run a test or even experiment with it right but if it's just a small piece you can it's easy for you to test it on your own computer it is also to spin up in a new for example cluster makes it easier for everybody to test it and and now what that does is changes the roles of people um I'm not saying that it's it was only because of microservices everybody's testing more but it's certainly a factor so it when you make you when you remove obstacles to testing you make it more accessible to everybody right so people feel more comfortable in in starting testing and the quality of the process of course improves what objects do you think of in particular that were removed so the first one is uh the automation side of it so for example to automate tools big processes for big applications the more complex is something is to automate uh you know the price people will do it because it's hardened and they will always do an ounce of cost and benefit and if the cost is like bigger than the benefit people usually don't tend to do it so when you remove the cost and and the value is also perceived as being higher we have people doing it more so one one is automation the other one is the complexity so when you have a very technical processes for somebody to run these things you make it only available for the more technical people so let's take out will not go into that right you'll see okay I'm not sure what is worth I'm not going to what it that far uh when you make it less less complex and less technical everybody will do it more so the experience has improved yes a lot how how do you think how do you see the experience improving for for example in for a particular you know projects for instance one example and one way we can observe this Improvement in testing is the release cycles of teams it's a common characteristic of the successful businesses uh that they are fast they are firstly in in delivery so the it right fast they have short release cycles and they are always like shipping stuff uh in these release Cycles if you don't want just them to be fast you also want them to have quality for you to make sure they have called you need to test it right you need to test what you just developed um so you can you don't spend so much time testing or you make testing like very easy and affordable so you you we can now see that most companies are very fast and then implementing like a lot of automation Tools around this testing um so it's observable that it became easier on on the cloud just looking at their release Cycles we've seen so much happen in the in the cloud native space what are you seeing as Trends then in kubernetes testing the the biggest trend is so we have in the last years people are adopting it right and um they they are learning uh all to exactly work with kubernetes you have companies that are running for a long time and they use the VMS in the past or they use all the Technologies and now this is they look at kubernetes okay this is uh great right let's use it so what they do is like they they start adopting it and they start changing their applications to live inside kubernetes so uh what we see there is okay they change that part of uh of deploying the applications and making them available to their um customers and as I said before the testing part they don't focus on that as fast right or they don't don't have that know-how on how to do it so the trends we see is always they start with kubernetes they depend applications there but then they don't test the applications where they live so they usually they they have the the old ways of testing and the trends we are seeing is that they are um adopting new methods of of running these tests they are using leveraging more and more kubernetes they are um automating things that uh way more they are running more things on the cloud things that can scale up and done as easily and they are spending way less time on things that they shouldn't care about right you shouldn't care about like the automation or to like plug out the system you just want your application to work and to be delivered fast to your customers right everything works fine everything um is released really fast and that's the main goal and the trend we are seeing is that there's a lot of Technologies on the ecosystem that are being developed with this in mind this how can we make the lives of developers and testers uh easier right how can we remove uh jobs from their plates testing has led to better approaches in terms of the developer experience one of the great issues that we have in the cloud and on-premise and everywhere is configuration uh you know and configuration errors do lead to terrible um outcomes and so what's the role of testing you're seeing emerging with helping reduce these configurations gifts allows you to see um that let's say you have a configuration issue right the problem is sometimes it will get visibility on it if you have the configuration of your application is like a bunch of different values different files different ways that you can run your application for you to detect that might not be you know easy for a human just to read the like the the files uh is not set soon enough so if you have automated testing to catch those things before you release to your customers is well when you catch this uh problems and so that's the biggest challenge is the complexity of automation there's always uh complexly around automating the test that your team creates right because the environments where they run they are different between each test they are different requirements or different things that they are testing so that there's like the uh the first part the second part is the it's hard to integrate testing tools because most testing tools were not made with kubernetes in mind or like to run in the cloud they are made with a specific purpose which is like testing their specific parts of the application for example if it's a tool for API testing the tool is very good at API testing but running in the cloud running kubernetes is not their main concern right the same thing with Cyprus for front-end tests or like any tool that test different aspects of the application um like most of them are not made for this right so there's like the one of the biggest challenges I would like to talk a little bit about how different people on the teams are are impacted by testing so I want to talk through you know people who are developers who are platform Engineers who are sres or even the the vice president or the CTO so let's start with the developers what are some the pain points you know there's yaml files there's overlays there's Helm charts and how all those things relate to each other how do developers avoid you know breaking you know these environments that's a very good question and uh you know most companies ask that which is very important the so everybody that contributes to the success of the application wants to know that it runs correctly right in the end of testing uh a test or a testing processes the process of um verifying if the application behaves as you expected it to right so you have some expectations of how it behaves you create a test to see if that expectations matches what is the the reality um so different roles in the company they contribute in different ways to the application right so the SRE just wants to make sure that things are running as they should the developers once want to see if the features that they just built you know do the functionality uh that they were asked to implement uh and testers are also focus on that so depending on of organizations not other all of organizations have the same rules or the same structure or the same processes so it changes company to company just just like to and we know that right each company is different and so all of them have a stake in testing right they want to test something but however based on their roles do you want to see different things um so the more we make that that testing process less complex or less are more easy to understand the better for the company right as you mentioned like the director the the CEO somebody just wants to see basically okay is the application is running like as Intel like all all the tests are passing um so that's like very important to them for sres maybe they just want to see like if the the availability up and down right the the developers want to see a specific feature so yeah yeah it's the requirements for testing changes between the roles so when it comes to the platform engineer you know one of the things you're really interested in is you know achieving that desired state right so I'm curious about the templates and the guard rails that platform engineers can put into place they quickly achieve that desired State how do they get started uh exactly so conference usually is don't ship anything to production that you did haven't tested so that's like one rule number one uh number two is perhaps of an environment yeah that very closely resembles production where you can run all of your tests so you can deploy all of your applications you should deploy data that's very similar um to what you see in production and then run the run like a lot of tests right to see if that environment is very close to production and the test pass right it's very likely that production won't fail as well right because they are similar that said the environments are never 100 similar right they are really just different things that you try to make one mimic the other so it's not 100 fail proof but it's a good measure of of knowing like being more confident on your deployments and make it like very fast so you shouldn't shouldn't make your developer theme weight for for let's say a manual QA that's going to test application manually to see if everything's all right before deploying it should deploy as as soon as you can as fast as you can um and we thought like everything everything should be automated and it should deploy um without waiting like for manual tests so as deployment as deployments increase there's you know always concerns about what might slow it down you know testing I expect you know in the inverse can be seen as speeding it up but I'm curious about get Ops and the adoption of get Ops for instance among from engineering teams how does testing affect that so it's like all key tops affected testing maybe it's like more the other way around so GitHub sees the for our watches that are not aware is the process of of starting the the states in your git repo in the Version Control y so instead of let's say you want to deploy application and you make a manual action you describe on your kids report that you want something to run and then something will make it be on that desired State as you have on GitHub so in testing is you describe all you want your applications to be tested in a declarative way and then you have some other tools that cons make a conciliation of what you want the things to to run and they will make it so and they're really um good way of these are these words is that you're the the code is also starting git right so if you also do the same way that you have declared like what CR code is and you also do the same things for your tests um it's very easy for you to make a a cluster that behaves the way you want it to and it's very easy to catch bags and easy to see how much it is wrong because if you make a change that makes your for example your testing process don't work so well that's in Gates right that's Version Control so you can see exactly what was the change that may have broken your system uh so that that definitely impacted uh testing we see more and more people just wanting to do everything on githubs because you know it's it's very easy to be fairly easy to implement um so yeah let's move on to sres you know they need to be able to quickly inspect and troubleshoot their cluster configurations they need to understand you know if things go wrong why they went wrong how they went wrong so comparing configurations becomes really important I'm curious about this and how testing helps in that process of comparing configuration exactly so SRA team is a team that needs to respond very quickly when something is done and they want to basically have be faster debugging as you mentioned right the the longer they spend time seeing what was the issue it means like there's more down time to their their customers right uh so that's very interesting and uh probably like the the people that want to run tests in kubernetes the most rsres because they really need to test live applications they really need to test it where the applications live and sometimes uh especially in critical systems the applications cannot be exposed to the internet for example right so if you run your tests outside of kubernetes and you are trying to test it things that are running inside right you you may you need to expose that to the outside and that probably won't work for all companies based on you know their their um their restrictions so uh especially for sres it's important that they run test inside kubernetes and they have like a lot of observability you know what's the status of their applications and finally let's move on to the vice president or that CTL who is really responsible for the business you know and they have to be thinking about delivery they have to thinking about how that application drives their business and they can't get dragged down by intricacies and you know and and the issues that come with configuration management and workflows what do you say to them about testing um well a good city you'll understand that testing is a important aspect right it shouldn't they shouldn't spend they should spend some time on it so on the chart term sometimes testing scene is okay it's a waste of time right but if you're if you catch a bag later on that you don't know what it is and it impacted your customers the cost to you is going to be bigger than if you spend some time testing it and you catch it sooner right so around the testing processes it's a steam testing like well when they are shipping like the the the biggest advice I could give them would see just just some sort of observe the process and then have a way of seeing or like what are the test results like if the things are passing failing so if you can just see that happening that will be the the best way to improve the quality of your product Bruno before we go tell us a little bit about Cube shop is a kubernetes company Cub either so they have six different projects that they created in the kubernetes area uh I'm the product leader of uh one of uh well the biggest the best product well I'm biased saying that because I'm the leader of this one but it is on this test tube uh it's a kubernetes native testing framework um so that's why I like I'm we are talking about testing and I know a lot about testing by talking with our users with a lot of Qi people so we really want to make the testing easy in kubernetes we want to solve all of the biggest challenges of of testing and so that's that's the the main goal of test Cube so cubeshop once really wants to improve the the space around kubernetes and test Cube wants to improve the the testing aspects of of the companies that have kubernetes clusters they want to test it they want to uh you know to to develop fast so you want to remove all those challenges for them Bruno thank you so much for your time I've enjoyed learning a bit more about testing so it's a pleasure being here if you like this video please give us a thumbs up and if you'd like to see more videos like this you can always subscribe to our YouTube channel we're on all the major social media platforms you can always find us at the newstack.io we hope to see you soon foreign [Music]