I appreciate you giving these tools a fair shake and telling us what you don’t like about them as well as what’s attractive. I think one of my biggest takeaways from your videos has been the utter breadth of options available to eng in this space right now.
That is indeed one of the main takeaways. We live in times with a higher frequency of innovation than ever before. However, that also means that making choices is harder than ever.
@@DevOpsToolkit I respectfully disagree that it’s hard. The options available to us right now are astounding! You can be as liberated or opinionated as you want. JX3 or tekton w flux or Argo or just Jenkins w tekton or or or. It isn’t difficult to make a decision at all right now based around an orgs current and short term needs. We’re on the cusp of the golden age of devops tooling in this regard, I’m just happy to be alive and working in an industry that is seeing such rapid QoL improvements for operators.
@@bobbyray8457 I agree with you on all those points except that "it isn’t difficult to make a decision at all right now." The more choices we have, the better, but also the harder it is to make a decision. It's easier to choose between 2 tools than 4, and it's easier to choose between 4 than 100. The more choices we have, the harder it is to choose. Now, I am not saying that is a bad thing. Quite the contrary. Choices are great. Nevertheless, the complexity of making a decision is always incremented with the number of choices. For example, if I focus on pipelines, it was easier to make a decision 10 years ago than now. Back then, it was Jenkins, and (almost) nothing else so the decision-making process lasted for approximately 25 milliseconds. Now, there is Jenkins, Jenkins X, CircleCI, Travis, Bamboo, TeamCity, Codefresh, Argo Workflows, Tekton, Drone, .... There are hundreds to choose from, and that's a good thing, even if it comes at the expense of making the decision process harder since competition drives innovation.
@@DevOpsToolkit yeah that makes sense for sure. I just feel like a team can easily isolate the direction they go with pretty early on (i.e. saas vs self, full pipeline management (jx, openshift) vs bring your own tools). I totally get what you’re saying, that too *many* options can muddy the water, I’d just counter that when you drill down to your specific needs and deployment options, it’s not as muddied as it might seem at first when you consider ALL of the options in this space. Edit to say this: there aren’t hundreds of options. Maybe a dozen tops especially if you don’t count commercial offerings like Travis. I really don’t think it’s as bad as you’re making it sound for an eng team that can discuss options properly.
For those of you using Kubernetes and with a preference to self-managed (not SaaS) pipeline/workflow engines... What do you prefer? Tekton, Argo Workflows, or something else?
I analyses Tekton two years ago, all your mention at end was my conclusions for a PoC, I prefer to use Codefresh, for this reason they are integrating ArgoCD. Also when I made a PoC of Anthos in same year the cost be very expensive.
@@WebMago It's always easier to use a higher-level tool like, as you mentioned, Codefresh. Tekton is more interesting either for software vendors interested in creating an abstraction on top of it, or, typically, bigger companies that have "special" needs that are best fulfilled by building their own solution.
Tekton newb but I’m interested how they may work together too. So far def on team Argo though. Tekton I’m trying to learn. I work for a bigger company so need to evaluate more.
@@stanrock8015 If you're referring to Argo Workflows, they do not work together since they are trying to accomplish the same objectives (k8s-native pipelines). If you meant Argo CD, it can easily work with Tekton in a very decoupled way. You just need to make sure that your Tekton pipelines are modifying the repo with the manifests and Argo CD will pool them and sync the states.
I had chosen between ArgoCD and Tekton and decided that Tekton will reach Argo features maybe in 2 years. Teton didn't have access control, and also couldn't use isolated namespaces that's why I couldn't compare them.
Thanks for the clear video For the relation between jobs and tasks, Red Hat openshift solves this strait a way with openshift pipeline using tkn underneath.
Hey! Great video. One suggestion I need from you. Which one do you think is the best open-source CI tool? I am using Jenkins because it is highly customizable and most imp I can check HTML reports in that. Is there any other CI tool available that provides the same feature?
Jenkins is battle tested pipeline tool so using it is a safe bet. On the other hand, you might want to explore newer alternatives like Argo Workflows or Tekton. They are likely going to be defacto standards in the future and adopted by other tools that will build on top of them.
The benefits are that they are k8s-native, meaning that we work with k8s resources and Kube API instead of directly with the API of the tool in question. Also, they tend to scale better, to be managed through operators, etc. Now, that does not mean that it is better than using something like GitLab CI or GitHub Actions. Those bring many benefits that Argo Workflows and Tekton do not. They are more user-friendly, have useful higher-level features, etc. Think of Tekton as a building block either for a team in a company to create a CI/CD platform for other teams, or as something used by other vendors to create a final solution. A comparison between Tekton and GitLab/GH/Bitbucket is not necessarily the right one. A better comparison would be between the tools that build on top of Tekton and those. So, it could be Jenkins X and OpenShift Pipelines on one hand (both are built on top of Tekton) and GitLab CI and GitHub Actions on the other.
Awesome video with beautiful insight :) . I wonder , how exactly is concourse different than tekton in this case ? Also, Why is it really special for K8's native ? Would this not suit well also for EKS maybe ?
I haven't used Concourse for a while, so I'll need to get up-to-date with it before comparing it with Tekton or anything else. In general, being k8s-native means that something was designed for k8s. Typically, that means that the resources are k8s CRDs, that it uses k8s API directly, that it provides k8s-friendly metrics and logs, etc. It does not necessarily mean that it is better or worse than other design choices, but that it was built to be an integral part of a k8s platform. As for EKS... EKS is Kubernetes that is partly managed by AWS. EKS is only a flavor of k8s so anything that applies to k8s applies to EKS.
If you don't mind and you have 5 minutes then can we meet on google meet for a quick call. I am thinking you are right person to talk about that.. so can we connet only for 5 minutes :)
@@DevOpsToolkit Could you also please explain diff between ArgoCD & Tekton? And also, could you please make a video on how to use both together & advantages/disadvantages of using them together?
Adding it to my TODO list... P.S. I made quite a few videos and articles related to Codefresh while I worked there. They were all published somewhere else since I'm doing my best to keep this channel as neutral and objective as it can be. Still, it might be a good idea to make a review with pros and cons on this channel now that I don't work there anymore (and can say whatever I think about it). TLDR; it's mostly positive.
Spinnaker is not in the same category with Tekton/ArgoCD. Spinnaker is just for CD. It work great but it's hard to install. Spinnaker cannot use OCI compatible Helm registries to pull Helm charts.
Sorry but the documentation is very poor. I don't know anybody can have achieved pushing images to private ECR repository, it is a crucial issue for a CD. Dockerhub is ok but AWS ECR is being used by too many organizations too. The documentation is missing too. Build ok, login ok but pushing is not clear and not explained. There is no any up-to-date working sample even on internet. Also, community is poor too. I opened an issue which is related the matter above, nobody hasn't answered for days although it's a very basic issue.
Tekton Is certainly not easy and I tend to rely on GitHub actions much more than tekton. That being said, I don't think that pushing images is directly related to Trkton but, rather, to the tool you use to build and push images. If, for example, that is done with Kaniko, the docs are there. Tekton is just a pipeline that orchestrates execution of other tools, just as any other pipeline.
+1 for tekton vs. argo comparison
I think that will be interesting.
+ 1!
I appreciate you giving these tools a fair shake and telling us what you don’t like about them as well as what’s attractive. I think one of my biggest takeaways from your videos has been the utter breadth of options available to eng in this space right now.
That is indeed one of the main takeaways. We live in times with a higher frequency of innovation than ever before. However, that also means that making choices is harder than ever.
@@DevOpsToolkit I respectfully disagree that it’s hard. The options available to us right now are astounding! You can be as liberated or opinionated as you want. JX3 or tekton w flux or Argo or just Jenkins w tekton or or or. It isn’t difficult to make a decision at all right now based around an orgs current and short term needs. We’re on the cusp of the golden age of devops tooling in this regard, I’m just happy to be alive and working in an industry that is seeing such rapid QoL improvements for operators.
@@bobbyray8457 I agree with you on all those points except that "it isn’t difficult to make a decision at all right now." The more choices we have, the better, but also the harder it is to make a decision. It's easier to choose between 2 tools than 4, and it's easier to choose between 4 than 100. The more choices we have, the harder it is to choose. Now, I am not saying that is a bad thing. Quite the contrary. Choices are great. Nevertheless, the complexity of making a decision is always incremented with the number of choices. For example, if I focus on pipelines, it was easier to make a decision 10 years ago than now. Back then, it was Jenkins, and (almost) nothing else so the decision-making process lasted for approximately 25 milliseconds. Now, there is Jenkins, Jenkins X, CircleCI, Travis, Bamboo, TeamCity, Codefresh, Argo Workflows, Tekton, Drone, .... There are hundreds to choose from, and that's a good thing, even if it comes at the expense of making the decision process harder since competition drives innovation.
@@DevOpsToolkit yeah that makes sense for sure. I just feel like a team can easily isolate the direction they go with pretty early on (i.e. saas vs self, full pipeline management (jx, openshift) vs bring your own tools). I totally get what you’re saying, that too *many* options can muddy the water, I’d just counter that when you drill down to your specific needs and deployment options, it’s not as muddied as it might seem at first when you consider ALL of the options in this space.
Edit to say this: there aren’t hundreds of options. Maybe a dozen tops especially if you don’t count commercial offerings like Travis. I really don’t think it’s as bad as you’re making it sound for an eng team that can discuss options properly.
Yes! A comparison video between Argo Workflows and Tekton.
It's coming :)
Nice presentation, enjoyed the sequential flow of explanation! Thanks.
another gem on this channel, thank you Viktor! 🙂
For those of you using Kubernetes and with a preference to self-managed (not SaaS) pipeline/workflow engines... What do you prefer? Tekton, Argo Workflows, or something else?
I analyses Tekton two years ago, all your mention at end was my conclusions for a PoC, I prefer to use Codefresh, for this reason they are integrating ArgoCD. Also when I made a PoC of Anthos in same year the cost be very expensive.
@@WebMago It's always easier to use a higher-level tool like, as you mentioned, Codefresh. Tekton is more interesting either for software vendors interested in creating an abstraction on top of it, or, typically, bigger companies that have "special" needs that are best fulfilled by building their own solution.
Tekton newb but I’m interested how they may work together too. So far def on team Argo though. Tekton I’m trying to learn. I work for a bigger company so need to evaluate more.
@@stanrock8015 If you're referring to Argo Workflows, they do not work together since they are trying to accomplish the same objectives (k8s-native pipelines). If you meant Argo CD, it can easily work with Tekton in a very decoupled way. You just need to make sure that your Tekton pipelines are modifying the repo with the manifests and Argo CD will pool them and sync the states.
Awesome review about Tekton! Wating to see your next video
I had chosen between ArgoCD and Tekton and decided that Tekton will reach Argo features maybe in 2 years. Teton didn't have access control, and also couldn't use isolated namespaces that's why I couldn't compare them.
Thanks for the clear video
For the relation between jobs and tasks, Red Hat openshift solves this strait a way with openshift pipeline using tkn underneath.
True. RedHat made the right choice incorporating Tekton as well as Argo CD into its platform.
I'm learning a ton.
Impatient for tekton vs. argo comparison :)
If everything goes as planned, I'll have it during the week after next
It's done and available in th-cam.com/video/dAUpAq7hfeA/w-d-xo.html
While I don't know much, Drone CI seems to also be a popular choice, if you are interested maybe check it out
Drone CI is indeed very interesting. I'll probably create a video about it as well.
Awesome as always
Hey! Great video. One suggestion I need from you. Which one do you think is the best open-source CI tool?
I am using Jenkins because it is highly customizable and most imp I can check HTML reports in that. Is there any other CI tool available that provides the same feature?
Jenkins is battle tested pipeline tool so using it is a safe bet. On the other hand, you might want to explore newer alternatives like Argo Workflows or Tekton. They are likely going to be defacto standards in the future and adopted by other tools that will build on top of them.
What is the benefit of such CI/CD pipelines over highly integrated pipelines of gitlab/bitbucket/gh?
The benefits are that they are k8s-native, meaning that we work with k8s resources and Kube API instead of directly with the API of the tool in question. Also, they tend to scale better, to be managed through operators, etc. Now, that does not mean that it is better than using something like GitLab CI or GitHub Actions. Those bring many benefits that Argo Workflows and Tekton do not. They are more user-friendly, have useful higher-level features, etc.
Think of Tekton as a building block either for a team in a company to create a CI/CD platform for other teams, or as something used by other vendors to create a final solution. A comparison between Tekton and GitLab/GH/Bitbucket is not necessarily the right one. A better comparison would be between the tools that build on top of Tekton and those. So, it could be Jenkins X and OpenShift Pipelines on one hand (both are built on top of Tekton) and GitLab CI and GitHub Actions on the other.
Awesome tutorial. We also extensively use Azure Devops for our CICD, its not k8s native, just a random CICD, but would love to see your take on it.
It's already on my TODO list :)
Awesome video with beautiful insight :) .
I wonder , how exactly is concourse different than tekton in this case ?
Also, Why is it really special for K8's native ? Would this not suit well also for EKS maybe ?
I haven't used Concourse for a while, so I'll need to get up-to-date with it before comparing it with Tekton or anything else.
In general, being k8s-native means that something was designed for k8s. Typically, that means that the resources are k8s CRDs, that it uses k8s API directly, that it provides k8s-friendly metrics and logs, etc. It does not necessarily mean that it is better or worse than other design choices, but that it was built to be an integral part of a k8s platform.
As for EKS... EKS is Kubernetes that is partly managed by AWS. EKS is only a flavor of k8s so anything that applies to k8s applies to EKS.
May I suggest Knative vs OpenFaas comparison ?
Adding to my TODO list... :)
awesome video, this video increase one more your subscriber :)
If you don't mind and you have 5 minutes then can we meet on google meet for a quick call. I am thinking you are right person to talk about that.. so can we connet only for 5 minutes :)
Sure. Ping me on Twitter or LinkedIn and I'll send you my calendly link.
Naming the task as well as the step as silly creates a bit of confusion at the beginning
Please do a video, comparing Argo Workflow and Tekton.
It's coming :)
@@DevOpsToolkit Could you also please explain diff between ArgoCD & Tekton? And also, could you please make a video on how to use both together & advantages/disadvantages of using them together?
@@chandup That's a good idea. I'll do my best to work on it ASAP.
YES! Compare. Add today! lol
I think it'll be published by Thursday next week (unless something unexpected happens).
Please make a tutorial about GitOps series learning ArgoCD FlexCD and Tekton CICD in Udemy
I will do my best, but I can't promise it right now.
Nice video, really interesting!
What about Codefresh ? :)
Adding it to my TODO list...
P.S. I made quite a few videos and articles related to Codefresh while I worked there. They were all published somewhere else since I'm doing my best to keep this channel as neutral and objective as it can be. Still, it might be a good idea to make a review with pros and cons on this channel now that I don't work there anymore (and can say whatever I think about it). TLDR; it's mostly positive.
spinnaker??
Spinnaker is not in the same category with Tekton/ArgoCD. Spinnaker is just for CD. It work great but it's hard to install. Spinnaker cannot use OCI compatible Helm registries to pull Helm charts.
Nice nice video
IBM cloud use it as a ci cd service.
True. Tekton is becoming the pipeline solution of choice for quite a few vendors.
Openshift uses Tekton
Oh yeah. RedHat is one of the fist companies to start investing in Tekton and is one of the main contributors to the project.
Sorry but the documentation is very poor.
I don't know anybody can have achieved pushing images to private ECR repository, it is a crucial issue for a CD. Dockerhub is ok but AWS ECR is being used by too many organizations too.
The documentation is missing too. Build ok, login ok but pushing is not clear and not explained. There is no any up-to-date working sample even on internet.
Also, community is poor too. I opened an issue which is related the matter above, nobody hasn't answered for days although it's a very basic issue.
Tekton Is certainly not easy and I tend to rely on GitHub actions much more than tekton. That being said, I don't think that pushing images is directly related to Trkton but, rather, to the tool you use to build and push images. If, for example, that is done with Kaniko, the docs are there. Tekton is just a pipeline that orchestrates execution of other tools, just as any other pipeline.
great video as always :)
how this compares to something like Drone.io? Ever used? thanks!
I used it. I'll add a review of Drone to my TODO and, at some later time, compare it with other pipeline tools.
+1 for tekton vs. argo comparison
+1 for tekton vs. argo comparison
It's coming :)