 Today, I think everyone got to be tired because it's just late today, so let's me walk through this fast and Let's me introduce myself. I'm a kind from Japan. I Super engineer in civilization and I'm just being deadly deadly. This is my kid and And let's take a break. I have some question. I think everyone is disaster. Let me so this would be super fast and simple and easy so I think it's proceed debris is things that everyone already know, but actually I hearing around Just the day and today, but everyone just talking about Argo CD not a role now that much. So I just get that I should asking again about what is progressive debris again and we all know that Why progressive debris exist is to about this dead end As an performance here, we all know that if so we go live the is really scary to be the one who on call and And so why? again for multi-cloud environment why multi-cloud environment and I think I can list up here a little bit Answer for that and it's maybe for diversity team and background or the service requires or it's just a legacy problem and In our environment, I'm just say my comedy Story we have about three department just under the company and Under each department. There's a lot of team and each team had the full control over the platform and they can choose which to you and basically with CI CD they use as well and What problem with that kind of environment? Basically, I can tell you like a story context like that in concept of a company one team Save instant team a and team B one use ZK E and one use EK add we can easily have work safety rating with Argo role now and as a platform team engineer, I Can manage this easily just one view one interface and one UI But how about when we starting get more cluster for TMA the workload for the platform team is increased then We have a new team called team C the team C is using Humanities as well, but this time they're not just using the ZK But they require for use the communities on premium as well So in that sky for Q and this multi cloud Flax city is quite famous for that. So they asked for supporting flux CD and the workload for the platform team is increased with Supporting flux CD Then the team B who use EKS Start thinking that they have something that not actually need to be a part of EKS. It's just Expensive and they starting you think like lambda and also for something like Running or something or boosting that are two other Adolescent service. They use lambda. So for the lambda code, they use GitHub accent or any CI for their CD and Also, we are team D who use cloud run. They are the server less from Google Clouds and they also use a detail action as the CD as well and Next time the platform team get more and more requests because this time team a and to see starting using terraform in in their Google load and as the as the infrastructure and goes for their cluster and More and more and team is is the old guy who just using ECS lambda and they use then keen for that And we can all know that there is a lot of workload for platform team if we have to What can I say many thing many different interface for many different guy of CDs and Something and something is not even CD. It's just CI and I Take this picture from my friend Whitney Yeah, I think this problem is just commonly because not just my company get this problem, but other as well The platform team have to handle multiple interface multiple tasks with different things so it's a lot of work for the platform team and Usually the platform team Remember is just too small compared to what they have to do So to survive this we think that we have to let other do our job so that we can Handle many things with a little bit number of resource and In business war we call is talk the duration So What do we do with that my set? We think that we need a way to implement the talk the creation idea and The second is the solution have to be easy enough so that other can Take over what we really get Deliver did it get to them and Also from my video friend She said that platform team so cheap the suit is the platform as their product And I'm totally agree with this thing That why we're starting a new product. I mean just Internal products is called by CD and Go back for the what do we do We need a way to implement the talk equation idea and with by city we do that It is the ideally architecture of a CD for a whole company that we think that could work for a whole company we have one control plane and Team can install the asian to their cluster individually and cluster pulling model to connect to the control plane and Go back with my company story and This time we can also see that the platform team workload is way Decreased because we only need to handle the control plane. We only need to operate the control plane Every team else just need to install the by the asian to their cluster and everything works To prove for my say it is the number in my company By CD control plane They use only one control planes and we can enable to serve the github progressive delivery for over three thousand and two hundred application and Yes, I don't understand that almost is the Q&A and For the second we need something that have consistent and easy enough interface and It is what we call the consistent and easy interface. It is a simple interface for configuration if you are using by CD to by CD as the progressive delivery tool for your application and oh Sorry, I'm just wrong copy one is go run and the other supposed to be cubanist this place it is the simple you eyes for control plane and for cloud run and for cubanist for cloud run. We have three state for primary chart disease one for most 10 percent then with uprover then 100 percent for most and for second things communities the calorie run out then great approval then primary run out then calorie clean It's a simple and just look the same for even different kind of application So wife's by CD. I think this is just a work take. So I trying to make it less work I will read it right away and By city is simple, unified and easy, but still powerful enough for you to create progressive by polite CDs and The best thing is that by CD is Provide you a same deployment interface to the application to any platform the third is that Is that not require you to install a any CRT you just only need to Prepare your manifest and prepare one pipeline definition then by the agent will take the rest Next one is the deployment create or does not need to be explore outside the cluster Next one is the building deployment analyst. We call it a da as a part of the pipeline So you can easily just you eat inside your pipeline and get Modemly progressive deliveries by pride Next one is easy to interact with any CI Also, we have an Inside metric Feature that you can review your deployment and last but not least The by see architecture is designed to serve thousand of growth application in multiple cloud For company scale, but still working out with small projects And I know it's still work take So I will walk you through some use case so we can just easily Images how by see look like in the rear you slide So we start with the common flow we can all understand that is the CI simple flow and With by city after we finish the CI flow We can trigger the CD and we go is the even watcher feature of by CD and In this picture I use github action at the CI But actually we can use any other CI just last step of the CI we Have some common like this and it's good trigger the by CD By CD when update the manifest repo we store your manifest Your ham your jam or whatever with the data that give from the CI pipeline and That manifest repo update green be reviews and approved by the developer again And at this point we will have an earlier set of what would be changed What wouldn't be? Regal side to your cluster We call it's plan preview feature. You can get it just like this in case of horizon use as your CI the we have isn't been print out thing like this one and After that by the asian wind pull the update of some manifest repo just like other github things and deploy it to your cluster and That will be whichever cluster so I Just want to mention a little bit about we could be deployed here. I just Tell you guys Leslie that by CD have and viewing a viewing analyst deployment State and we call it idea and you can conflict it like this basically in this In this laugh I use the Prometheus query I Use the prom query with data from prometheus. It could be whatever matrix server You can set up it by yourself and the pipe the asian green load it and You can win right query like this and check it out if any problem happen the analyst state win before and By the wind run by jaw deployment right away So for the uncommon flow This time the story a little bit changed This time your poor regret is directly update the manifest repo and the Bible II will notice the chain and It's my chain on the teleform things Then after that we introduce thing called posting which is the trigger after the application we Deploy the previous application is teleform, but it's can trigger the deadline for the QITS application as well and To do that You can just make the conference and pipeline like this. Just simply mention the application you want to trigger after the pipeline is finished for the first application and It's not just one next but the next and the next and the next as well just everything is the config like the both I just mentioned So that is not Just that we have a lot more feature that I think develop a good love to have and We have an official site here. You can just Scan and walk around. Maybe if you feel interest you can Go for the play side as well. That is the live demo and we also have a free trial or request and you guys can Ride to the form and we will generate some Project for you so that you can use our set control plane to use to display your application and The current status of 5 CD the latest version of 5 CD is 0 by 45 45 by 3 It's already brought some use wisely in our company. I just said you previously is handle over 3000 application in our company and 5c already enjoy CnCF Sunbao from this year and For the roadmap of the project we are the middle of the way To reach the version 1.0 with several big thing Chasing the first thing we are plan to change is we want to make Adopt for the blockable architecture so that the 5pd Can support more application currently we only support 5 chi is communities terraform Is yet lambda and current the second is we are trying to find a way to make the our wall complexity or Open-cruise thrown out work with 5cd so that 5cd can remain as the CD pipeline and we could be used as the app layer can be freely changed and The test is we are trying to add more feature on the pre and post deployment Save instant We know that just dbt and open AI is quite widely used today So we're thinking that some kind of look or print or print review feature can be What's up more power if you can use power of AI as well So it's still a fresh project and had a lot more room to grow so if you guys have the same problem and You feel that 5cd can work with your case Feel free to Bing earth on cncf slide. We are in 5 cd channel the Officer site is 5c.f. And our github report is 5cd slash 5cd time for listening. I think that's all Any question? Okay, so I think we can stop it here Good question Is your product a SAS product specifically? Yes Just wanted to make sure sure if you feel interest on open sourcing, please just come for Just like on cncf. You're very well there. Thank you Thank you, Ned. Thank you so much. Sorry. Do have one question How do you guys handle like rollbacks in terms because let's say pipe cd handles both Obviously handle both progressive deployment plus the infrastructure side as well like terraform everything. Yes How do you guys think about rolling back, right? So if there is a deployment failure, you have analysis template In the case of a failure Do you roll back everything including the info update or just the application itself? For the terraform thing, right? Basically we use We do follow github. So if anything happened the terraform will apply the previous state That could be I mean on the github On the github flow. I mean the previous Commit is the stable one. So we try to apply is first and if not everything go right with the terraform That time the the platform team will get in and fix this for that. I mean For the terraform is the only thing that we can do for now right, so is the Analysis bit focus can be like chucking to anything like in between terraform update as well as the application update basically the basically Trying to make the same interface for I mean as many as possible. So the wrong the wrong bike The wrong bike term is work easier on application things You are so right and for the infrastructure things the terraform even if you want to reapply the ladies Commit that something may cannot even run back understand that so we just trying to make that grown back able UI on the UI which means that at least we trying to make the terraform apply for the latest Stable commit and if it's go right then it's go right and if something happened we can just do it later Cool. Thank you. Thank you