GitHub Demo Days, 16 Jul 2021

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: It works in dev, test and staging, why not in Prod? 😡 How Trilio uses GitHub Actions #DemoDays

Description

2:04 - Program Start
3:49 - Overview
10:02 - Multi-Cloud Data Protection and Management
14:24 - A persona based view of the challenge
18:14 - TVK Runner for GitHub Actions
21:54 - live demo

GitHub Solutions Architect, Andrew Mccoy (Moose0621), is joined by Prashanto Kochavara and Ben Morrison of trilio.io to dig into Kubernetes, cloud-native deployment pain points and how they use GitHub Actions along with the TVK to solve issues when deploying to Production.

https://www.trilio.io/triliovault-for-kubernetes/

For upcoming Demo Days: https://www.linkedin.com/company/github

A

Hello: everyone, hello, linkedin, hello, devs, hello, folks on linkedin good morning, good afternoon, good evening, good night, always nice to see worldwide representation. Today, uh today we're going to talk about a particular cloud, late native deployment pain point here on demo days and uh to help me with this topic, I'd like to introduce you to moose, uh who is a field solutions engineer over here at github, and we also have some additional special guests from trillio.

A

I've got uh preshanto and uh ben roshanto is the uh director of product over at trulia and then is the solutions.

A

If you could take it away.

B

Awesome, thank you so much aj and actually yeah. I gotta update my linkedin, I'm actually now um on our uh advanced security team over here at github as a field architect, um so I got gotta go and update that so happy to be here today with my friends, ben and pk.

B

um Interestingly enough, I got to give a shout out to uh rob and dalton menhall over at um trillio for helping get this together through through our channels, uh working together in the past, so um been got connected a couple months ago, just kind of understanding. You know what what trilio could do, potentially not just with github but for the github community as a whole. um So I'm happy to have you join them, um be joining them.

B

uh That's why we do it live right, uh be joined by them today, so they can kind of show you uh what they've uh has come out of conversations. They built a really awesome, github action, um they think is going to provide a lot of value to the community here. uh Pk.

C

Ben pk, I think you're still on mute there. So.

D

The moose is absolutely right, you know uh the joint collaboration between github and twilio has been extremely uh valuable. You know, we definitely feel we have built something really intricate over here, which is going to solve the challenges that we experience. uh You know in a cloud native uh delivery model and uh ben, and I are going to be super happy to present everything uh you know show you all a demo as to how it works. uh You know what everything is built and the value that it's providing in the end.

D

Okay, so without much to do I'm going to put this in presentation mode and we can get started so uh from an agenda perspective. uh What I'm going to do first, is I'm going to talk about the challenge or the problem that we are solving over here? uh Next, uh we will get into the cloud native data protection platform of trillio and uh we look at the solution uh as to what is the technology behind the solution to the challenge and how it's powering it.

D

After that, we will uh provide a quick uh introduction to the tbk github action, uh the the workflow or the github action that trillio has developed uh and then ben will be going into a demo, showing you exactly how everything works and how we are going to solve that pain point for you and then finally, we'll wrap it up with a conclusion, and if we have any questions we'll try to answer those as well.

D

Okay, so one of the most common ci cd issues that we hear about you know when we talk to our customers, uh is that you know when they are developing code. You know everything works well in their dev environment, they move it into their test environment. They are able to test properly, and you know everything works there as well. Staging wise. You know. Just before launching into production everything looks clean and when you launch into production you hit a bunch of different issues and then oftentimes.

D

You end up looking like that image on the bottom right. You are thinking about hey. How did this happen? Why did I miss this and when you actually find the root cause, you feel that hey? This is something that I should have uh figured out earlier on, or something that I could have stopped from failing in production now, one of the biggest reasons why this ends up happening is because, in all the other silos that we have of you, know dev test staging environments.

D

We end up uh generally testing with older copies of the data or you know it's stale data, or you know random data, which is not a true representation of your production data and, as a result of that, you know, issues that happen generally more during runtime you're not able to address those or surface tools uh before launching into production.

D

So the need to test with production data becomes very, very important now in.

C

D

To keep our uh conversation and this uh the session little interactive, we have a poll question based on the problem statement that uh I've just described and what we want you to do is use uh the linkedin live chat to type in uh your answer. So the question here is: do you face these ci cd? Challenges of you know, passing every step until production, but failing in production and your options are? Yes, you face these challenges, but you do not have a solution.

D

Yes, you do face these challenges, but you have a solution and then c is obviously you uh have this challenge, but it's really not important to you again. We are looking for your responses in the uh linkedin live chat and uh uh just enter the letter corresponding to your answer and uh aj. I'm gonna look to you because I cannot chat. uh You know.

B

I'm keeping an eye on it too, pk perfect, so we're it's just actually coming out now I can see it. Okay,.

D

So I'll just give them another 30 seconds to 45 seconds for folks to look at the question and then quickly enter some an answer into the chat.

B

You know this is definitely bringing me back to my days years ago. As a devops engineer, I wish yeah definitely face a lot of these challenges. This misconfiguration of files password rolled, got updated. um There was control, ms in the uh configuration popular one as well. Definitely.

D

Yes, yes- and I think uh you know, the cloud native journey is definitely helping solve some of these problems and you know we'll definitely be kind of addressing those right now.

B

Definitely a good uh smattering of answers here. um Looking for the trend, these definitely a good amount of bees.

B

But definitely a good amount of a's nuts not seeing any c's really okay, so definitely.

D

This is someone saying sorry, okay, I guess the general consensus, though, is that this is definitely a problem, and uh you know you may be trying to solve it in different ways uh again and today we are going to be talking about how you solve this. In a cloud native, uh you know, kubernetes native environment.

D

uh You know making sure that your application delivery is as fast as application development, uh so we'll jump into talking about the technology that is trillio trillio want, which is powering the solution and the github action that we have built so uh trilio as an organization uh you know was founded in uh 2013. uh We have. We are already the leading data protection data management solution, for uh you know openstack and radar, virtualization environments and now vsphere having the kubernetes uh the kubernetes journey as well.

D

So what you see in front of you uh is the trillionvolt management console and uh the way we have built our product is: we've made it completely uh kubernetes native. So as long as you have a kubernetes uh qualified environments or cncf conformant kubernetes distribution, trillio will work out of the box uh in it and whatever we are showing you as part of the github actions that we've developed, they will also be supported without any additional configuration that is required.

D

Secondly, what we've done is in order to communicate with storage uh or your application volumes or your persistent volumes. We leverage the container storage interface uh so that you know we can keep that data layer extremely agnostic. So even if you're dealing with heterogeneous environments throughout your you know, development, your test clusters or your staging uh clusters. As I mean, if they are completely heterogeneous, you know trillio will still work out of the box on them without any configuration required.

D

And obviously we know everyone is following a you know, multi cloud or a hybrid cloud strategy, so you will be you know, running the software or the solution in multiple places and trillio has you covered for that as well and then? Finally, what the technology supports is uh supporting your. We are completely application centric. So what we allow you to do is capture your entire namespaces.

D

If you want to uh capture specific helm, applications uh capture operators themselves, like you know, capture the operator, objects, the operator, custom resources and the application that the operator is managing all of that as one unit and then finally, we allow you to capture whatever you want within your kubernetes cluster via labels as well. So, overall, you know, 360 degrees uh makes it a really nimble, flexible solution working in any of your. uh You know uh deployment services pipeline, uh however, it may be constructed. Trillio can support you in there now.

D

What is it that trello does you know? We spoke about hey that these are the features that is capable of you know, being distribution, agnostic, storage, dynostic, but what is it that we actually provide? And how is it that we enable that protection and recovery?

D

So at the very first level you know from a kubernetes cluster at any name space level we can capture all the metadata components and the data components, and we can put that in a centralized repository which could be nfs based or s3 based.

D

So we capture all the metadata the data volumes and we store it in that central report now, once it is in that central repo, you have the ability to take that same data and restore it back into that same kubernetes cluster. Obviously, you can overwrite what you had in your namespace, but you can also deploy it into another namespace as well, and what it also allows you to do is take that application that you captured and move that into another namespace of a completely separate or completely different, kubernetes cluster.

D

So now you can visualize that this method of doing point in time captures and point in time restores trillio can. Actually you know between your production environments and your dev and test environments. We can easily capture your running application and, along with the metadata and the data aspects of it, bring it back into a secondary environment where you're performing your testing and validations.

D

Now I think it becomes very important to understand the different players within the kubernetes system and how they would be using this solution of trillionvolt, as well as the github actions that we have developed as part of it.

D

So on the extreme left we can think of the persona of lisa. Lisa is the developer. Who is you know, uh 80 of the time writing code, uh building, microservices and focused on you know, developing the application and her main challenge again is to ensure that she can test with production data so that she can, you know, enable or increase the success of launching, or you know, delivering software back to customers.

D

Now lisa works very tightly with brian uh brian is the sre. uh He works with lisa to take the microservices that she has built, run it into the run it in the kubernetes system and making sure that it's uh you know it has uptime. It's continuously running. You know from a monitoring logging perspective, you are making sure that it is providing all those metrics and stats and again, the main thing that brian is also concerned about is making sure that uh you know when he launches this application in production.

D

It is successful and he doesn't have to keep re-tweaking and rehashing and fixing issues that could have been caught beforehand and as we move from the development side to the upside, we have rob. Rob is also, uh you know, think of him, like an sre from the op standpoint, who is building all these clusters for the dev teams to play in and develop their code and run their applications in now. Rob's intent is also he's looking at everything at a macro level right.

D

He he is looking at all these name spaces together and he wants that easy button or that easy workflow to restore everything in case a disaster happens so he's thinking more in terms of disaster recovery.

D

You know, obviously, if any kind of corruptions or security breaches happen around ransomware recovering from those those are the different activities that rob is looking for and then finally, you know looking or going deeper into the upside. We have jane jane. You can imagine her to be the uh kind of like the ide director who is focused on cost uh related items and focused on business continuity because of which migrations of applications become very important and year and year.

D

So overall, this is how the personas, within a kubernetes system interact with each other and obviously what they do. Is they leverage a lot of infrastructure as code right, their infrastructure as code is, uh you know, placed within the git repository itself, so in github, and they not only have you know, lisa's development code over there, brian's deployment code and rob's deployment code for deploying the clusters, but also the application and how it's built.

D

Everything is kept within kit and obviously there are paradigms like get ops that lisa and brian are using to deploy everything on the applications into the cluster.

D

So our focus for today's conversation and what we have built, the tvk runner for, is going to focus on the day, zero aspect of brian and lisa working together and ensuring that they are able to deliver software successfully and with as minimal of issues.

A

D

Possible, okay, with that, I'm going to start talking about the github actions, runner that trillio has built. So I'm going to start off from the left hand, side that is the developer laptop. You can imagine someone writing code and checking it into github. Now within github you have. The concepts of workflows.

D

Workflows, as the name suggests, is an orchestrated pipeline of performing certain tasks and as part of the workflow, you leverage the runner that trillio has built. Now. This runner basically has a source configuration and a destination or a target configuration.

D

What that runner basically does is once you check in code, first, a backup job gets triggered and as part of that backup job, the tvk runner will go to the production cluster, which is the source cluster capture, everything from there. So, let's suppose you had a mysql database running in production. We will capture all the metadata and the data objects, but get it into the centralized repository.

D

Then the second part of the workflow would be the restore job and what happens in this stage? The runner again takes that copy of data that you had captured and brings that into the destination cluster.

D

So what happens is as soon as your code has checked in on one side, you are running your integration tests, uh making sure that you know all the code can be packaged into.

D

uh You know a container image, and while that is happening, you are also going fetching your production data and bringing that into your test environment so that once your ci tests pass, it's ready to uh run against that newer copy of data that you have just brought in okay um now there are obviously other items also that trillio provides as features so, for example, you know if you need to anonymize the data from the production side to the destination side.

D

Trellio provides a lot of hooks which can be injected into the runner itself, so that you can talk to your mysql application, your anonymize a column of the table. If you need to and then bring it into your dev and test environments, so all those granular capabilities are also available.

D

Lastly, the uh you know when you have multiple clusters, it's always possible that you are running different storage classes or different storage types in your production environment versus your test environment. What this runner also allows you to do is on the fly, as it is moving that production data from uh the source side into the destination side. We also allow you to change the storage class on the flywheel and category store.

D

So again you know talking about that heterogeneous aspect. You know fully agnostic, no matter what kind of environments you're coming from you will be able to do this. You know, day in day out without any issues.

D

Okay, so now uh I'm going to hand it over to ben to basically put the rubber to the road and whatever I have spoken, show that in action as part of a demo.

D

And then, over to you now.

C

Okay, thank you for shanto.

C

Can you uh see my screen in the lens terminal up here?

C

D

C

Good to go sounds good, so we're going to do a demo today, showing a backup and a restore from one cluster into another cluster using that github runner, and so we have two clusters here. Our first one is going to be a rancher cluster and this is going to be the source cluster and then, secondly, we have our gcp ocq cluster. This is ocp running on gcp for our target or destination destination cluster.

C

So, first and foremost, as preshanto said, our github runner is running in our source cluster. So you can see this example runner. We have here using our repository, trilio demo, flux demo and it is up and running the application. We're going to be migrating over from one cluster to another. Today backing up and restoring it's going to be a my sequel application.

C

So here you can see this was deployed about 17 days ago up and running one out of one all is good. We have three different front, end pods and a back end. Mysql pod running. We have two different services here for the front end and the my sql.

C

We have two different deployments for that front and back end, and then we have these two replicas lengths all three out of three and one out of one good to go.

C

We also have here a pvc associated with this mysql database, we're bound and good to go and you'll, see that it's using the storage class of ebs, storage, class, ebs, sc and then lastly, we're going to show here that there are no backups that have been taken yet of this application, the trulia runner once it starts its workflow, is going to take a backup here in the source cluster and then restore it into our destination destination. Cluster.

C

This gcp cluster here so now is to show you that none of these that the mysql application is not running here in this space. As long as I did my homework correctly and make sure I removed it, we'll see no resources in this.

C

Namespace there we go there'll also be no pvcs.

C

And there will be no restore that has occurred yet in this destination cluster.

C

Now switching over we're in github here and we're looking at that trilio demo flux demo, repo, where our workflow is sitting to use that github trilio runner, and so we have a few different files. We're going to look at here. First, we'll look at this source key config, as you can see here, it's that same rke cluster, that I showed you before for the source cluster and we just have our keep config in here. So the runner knows which, uh which cluster to be pulling from.

C

We then have our destination cube. Config. You can see here ben gcp-4, that same gcp, ocp cluster, that we were using before for the destination that had none of that mysql application running in it.

C

And then we also have our environment variables here. This is where we're going to see what target we're going to be using to back up that back up to store that back up to and where to pull that back up from. In the case of the restore, we also have the name space where that target is sitting on this uh source cluster here and then the backup namespace, where the backup is going to occur on that source namespace, which is that bend demo, namespace and then the destination cluster.

C

We have again the target we're going to be using the target namespace on the destination cluster, which is demo restore and then the destination namespace on the destination cluster, which is demo restore.

C

Now we're going to go into the details of the workflow of our runner here, you can see we're looking at the tvk backup restore runner.

C

We're gonna have this configured so that it triggers the workflow every time a push or pull request happens on the main branch. Only we're also going to make sure for the case of this demo, especially that we can uh manually execute that action if we want our jobs are then going to be in the case of a backup job first, as preshanto alluded to, and then we're going to do a restore next.

C

Here's a few checks and steps that's going to go through as it starts that backup process making sure cute cuddle is installed, pulling those environmental variables.

C

We're then going to create the manifests of our backups themselves and you'll see those will be in the runner, backup plan yaml file and that backup plan will be named runner, backup plan.

C

We're then going to create the manifest for the backup itself, runner, backup, yaml and the name of this backup is going to be runner, backup and we'll see that once the action is complete back in our source cluster, we're then going to make sure that we make sure that we wait for any other existing backup to be completed before we start this backup, then, from there we're going to execute the backup plan and backup itself create them and thus kicking off the backup.

C

Now, the second part of our action, our second job here, the restore job.

C

We're then going to do a few checks again make sure q pill is installed.

C

These environmental variables pull that from our dots, enb file and then we're going to create the manifest for our restore- and one thing I want to point out here- is this storage class transform since we're restoring using trilio from one cluster an a rancher cluster into a secondary one? A ocp cluster running on gcp they're, two completely different distributions.

C

So we need to make sure there's some consistency there when we do our restore, so this transform component is actually going to go into the backup and change the storage class before we had ebs sc for the storage class of our backup, and this new application is going to be running standard csi for its storage class.

C

We're going to check the manifest to make sure it's good to go. We're going to wait for that initial backup to be completed, that first job and then from there. Finally, we create the restore itself and the restore gets off and running.

C

Now I'm going to go into the actual action section, go down to my tbk backup, restore and go ahead and kick off this workflow in just a minute here. You'll show it pop up on the queue.

C

And here you'll see some details that the job is currently cued. It's going to show us what that total duration is once completed, and you can see here the two different jobs we have. First, our backup is going to occur and then secondly, it's going to kick off our restore.

C

So here's that series of checks that we talked about before in the uh workflow itself making sure cute cuddles installed flowing the environmental variable.

C

And then now is going to kick off, creating that backup plan and backup now preshanto, as I mentioned before, the the backup itself in this demo is going to take about two and a half minutes. Maybe three minutes and the restore itself will also take about two and a half to three minutes so at this time, if there are any questions in the chat or anything you want to address for shanto I'll, uh let you answer some questions. There.

D

Sure awesome man, thank you for driving us through that and uh yeah. We can see that the backup plan and the backup is getting created. So basically this would mean that uh you know the production data is getting captured now and we are moving that into that. Centralized repository.

B

um I just want that's some great modularity in there just kind of thinking how the community can use that, like just being able. I love that anonymization piece you were talking about, because that was always a big struggle. I had back in your back a few years ago now at this point um with kind of you know, you don't need customer data or any kind of identifiable information, not just no there's some other regulation around that too. So that's that's.

B

A really big um help um as far as making sure you're you're treating data as it needs to be treated too through your workflows, correct.

D

And then um you know, that is definitely one intricate piece and, as we were talking about the transforms you know, storage is one item that uh you know. Customers would definitely want to massage and tweak as they're moving the applications around and then you know these transforms are so powerful that anything and everything within your yaml file. You can technically tweak that as part of the overall operation.

D

So in this runner that we have created, you know we have provided some uh injections for the storage class, provided some uh uh integrations for you know changing the service port and things like that, but users can actually you know we have an entire api library of additional features and they can actually use that entire api library to inject any kind of transform any additional features or things that they want to do as part of the capture and restore process.

C

To me for shout out, yeah really gets into how customizable the whole process is. Since we integrate natively into that kubernetes api as well. You can really use the trilio tool to to build and change your backups, your resource, exactly how you want, especially.

D

An example of those.

C

Transforms you just said: okay, so there it looks like our backup job completed, and so now you can see the restore is kicked off. It's just gonna wait here for a minute for the job to get picked up almost looking for that.

C

C

That looks like one through a series of checks.

C

Checking to make sure that restore manifest is good.

D

Okay, good, that's great flowing very well, and now, while we were doing this, uh there is a question I see in the chat window. uh Do you have protected customer data in your production data that you are copying over to a test environment? So uh ukachi? Yes?

D

So, as I was saying earlier, when we are doing the capture, we have this feature known as hooks and as part of hooks, you can actually start talking to the application directly and you know make sure that when you are doing a copy of that application, you can anonymize a particular table before that capture happens. So if there are any security and compliance reasons where you cannot bring production data back into your own environment, your customer sensitive information you can manage and handle that very easily with tradio.

D

Okay and then I think the follow-up question is, if you have a way to anonymize customer data, then why not create synthetic test data? That is customer alike?

D

Yes, I mean there is there is the ability on you know you could technically do that, but the the point that we are trying to uh build on over here is that you know your production environment is uh continuously changing. There is new data that gets added new type of data, new data forms that are getting added, that same data is being used by different microservices and when you are injecting new variables into your code, you know, maybe it may be a floating point. Integer double pointed here.

D

Those type casting issues are something that you know you can easily capture. If you are constantly you know, taking that production data and testing with it.

B

A

D

Definitely makes.

B

Sense, it could be also, you know, situational dependence. Sometimes you can create synthetic test data, but it I it sounds like this is a really great way to continue as production data does change, making sure that you're continuing to test against that that those same data sets um so that you capture those air um defects. Further upstream I mean that's, it's not just something here, that's something I think we can all relate with there um couple other questions I can um check. How easy is this to deploy from the chat from david.

D

um So in terms of deploying the entire trillion wall solution, uh assuming you have a kubernetes cluster and you can run a helm command, it should take you about five minutes uh to deploy, configure and get your first backup run.

B

Well, that's yeah! That's that's turnkey! I like that um another one from audrius uh what's happening when production debt is not 20 gigabytes, but a lot more yeah I've dealt with some customers. Lately they told me that you know some models. They create can be upwards of 20 petabytes of data. um So.

D

Right so one thing that we do is uh you know we provide incremental captures as well. So let's say you uh did this so the first time you got your production data in and all your incremental testing activities can be based on the incremental captures, which will be overlaid on that base image that we already have.

D

um So you know that way. You can continuously uh ensure that your data is, you know current, and you know you're testing, even however big your data size may get. You are still able to test with that, and this is going to be a key feature that we provide as part of the product and a few upcoming months.

C

If I can chime in here now as I'll show you about restore it uh completed, do you want me to go ahead and finish up and show the pods running the new cluster yeah? So let's do that great, so here, first off we're in the source cluster again that rancher cluster in the beginning, I'm just showing here that now we have that runner backup that has been completed. It was just done the past couple minutes here. It was a namespace backup and restore that's how we chose uh to do it.

C

This time was a name, space, backup and restore, and we're available and um jessica said before there are incremental backups. You can have to save space on your storage target, but here we did a full backup now going into our destination cluster.

C

We saw that there were no mice, equal objects running in this cluster before go ahead and take another look, and now you can see all four of those pods are up and running our three front, end pods and our back end, my sequel. When I have one good to go, just created the past three minutes there we have our two services, our two deployments and our two replica sets.

C

We also have that persistent volume fade in here bound good to go just create about five minutes ago, and here you'll see, as I point out before, we switched from using one c one storage class in the other cluster to now using uh the storage class that is used in this cluster, that standard csi and once it was all in that restore manifest, that transform was um was inputted into there.

C

There was nothing you saw that we had to do manually on our end and then, lastly, just to show you we have the restore itself up here. Runner restore just as it was named in the uh in the workflow, is completed just done in the past couple minutes here, and it was that namespace restored and with that, uh I'm I'm good with the demo and I'll toss it back over to you. Prashantha awesome.

D

Thank you so much ben. I think uh you know that was extremely valuable. um You know especially doing it live. uh The demogods were definitely with us, and uh you know uh I think it was presented in a very succinct fashion. You know showing exactly what we spoke about in theory and you corroborating it in practice.

D

um You know, I'm pretty sure it's going to be very valuable for our audience here uh we can move back to my slides now and put this back into presentation mode. So, in conclusion, what uh what we want to say is that you know we all know. Kubernetes is equal to all your micro services plus infrastructure escort right. You want to make sure that you're managing your kubernetes environments via infrastructure score and not manually, making changes, that's the recipe for success and that's how you will be able to streamline sort of operations.

D

Kubernetes focuses on application, development and application delivery, and you need to ensure that you're using kubernetes in this intelligent way. You know how we were showing testing the production data automating, the entire workflows. You need to employ deploy these practices within your own environment to get the real benefits of a cloud native or kubernetes environment.

D

We've spoken about trailer world being a purpose-built solution, uh managing your uh data and your cloud native applications together as one unit uh github actions. As we all know, our purpose build tasks supporting you know, ci cd pipelines and what trillio and github actions together are providing you are: is the capability to continually test with production data, to increase the software quality and delivery process for your customers?

D

We are all aware of you know, backup and recovery being used for day two activities. You know disaster recovery, migrations and things like that now today we have shown you how same tool. You know why our infrastructures, code, environment, github actions can be used as part of your ci cd or your devops processes.

D

What we have shown you today can obviously be further extended. You know we have been hearing about devsecops. You can think about dev resiliency ops. As you are deploying your applications in environments. You want to make sure that they are protected as they are getting deployed as well. So with the power of infrastructure as code, you know power of trillio and github actions.

D

You can make that possible as well, so that your application is protected right from the time it lands into production and you're, not really waiting for your operations teams to focus on those items.

D

So with that, uh we conclude our presentation. uh You know we definitely uh thank you all for your time and paying attention to what trilio had to present. um You know we will be available to answer questions right after and if there are any questions, even after the session you can uh reach out to us, uh my name is shanto. My last name is cochiwara, so prashant without coach of our I o and then dot polishing at trillium io as well.

B

Looks like uh we do have a question for you, pk and ben I'm ravi. How do we ensure application consistency? Is that via done via hux,.

D

Yes, so uh via hooks uh is the exact way to get application consistent, backups. uh What we have done ourselves is. We have qualified a bunch of different uh databases like mysql postgres. uh You know cassandra, you name it all. These different databases, they've been qualified validated and we have provided the exact ammo files and everything that you would need to use uh directly in get books, which is our public documentation.

B

Oh perfect, so, just like you know, an action marketplace will do where yeah excuse me, where they'll likely be able to find um this can have action, um though they can pull that information together for themselves.

D

That is good yeah, so one of the things we are doing with uh whatever ben has provided and spoken about and the demo he did. We will be working with. uh You know the github team to host it as a solution that customers can leverage directly from github. We will host it internally as well and we will continually be you know, updating it with additional features and providing a own custom module that we built. But you know, obviously uh you know, developers and sres are free to build their own modules, leveraging the trillium solution.

B

Now um it's an open, I'm guessing the actions in open source repositories any if anybody makes an enhancement or interesting things like that. Pull requests are welcome on that exactly.

D

We will uh I I we have it right now in a private repository, but we will be exposing that we will be making some noise around that evangelizing that but, yes, the end goal is to make sure that it's an open source model so that you know we provide the starting point and our users can keep enhancing that, and you know have that community involvement to make sure that life is simple and easy.

B

Awesome sounds great, don't see any more questions in there. um No, this has been great, um definitely been enjoying uh working with you, the team so far today, and definitely this this solves a myriad of uh questions. You know I I now deal primarily with our advanced security stuff, but really helping folks as they're rolling out get up actions. You know this is really that next generation level of problem folks are are now and now they have a best and bright solution to do that. Right, native on github actions, this has been great yeah, absolutely.

D

Thank you. So thank you for the kind words and you know I'm pretty sure you know there will be additional innovative methodologies that you know we'll all come up with earlier we'll be coming up with, and you know we'll be more than happy to work with. You know the excellent team at github and you know, add more value to you, know the ide world in general.

D

Awesome. Thank you all. Thank you. Everyone. Thank you. Moose thanks, moose take care.

D

D

A

A

A