Microservices In the Cloud Native World

Key Takeaways

hello welcome to the news tech makers podcast where we talk about a descale application development deployment and management thanks for joining us today one buzzword that I mean it's not a new thing but me here is micro services mm-hmm what is it Oh micro service if you it's best easiest to understand mic reserves in the context of the previous architecture the the client-server era which was monolithic three-tier so a style applications the front end the middle to your business logic and then back in data store like a database in those kinds of architectures the front-end actually consists of many different types of functionality the middle tier pre consists of many type different types of functionality there might be something that is transaction processing with the back end something else that is so talking to a customer database and when you've got a monolithic application what you saw was slow release cycles which were in match with the kind of IT infrastructure that everybody was using that when you go to the agile world that the cloud offers which is on-demand instant creation of compute resources that old model breaks down and what you've got is a mismatch between the delivery model of the application architecture and what the underlying platform can provide so micro services have come along as a way to break down the monolith into its constituent pieces and have different developers maybe even teams contribute and develop those and then release them independently of every of the other teams that are working on the other micro services of the larger application and that provides this level of agility that in the application architecture that can match the agility benefits that the underlying infrastructure can provide and the fact is that in many complex applications if you're not doing that then you're not going to be competitive so that this micro services approach has been driven both my trying to take advantage of the infrastructure as well as to remain competitive and that doesn't mean it's the answer in all cases but certainly in complex projects that have multiple teams or different pieces evolving at different rates or scaling differently then it's very effective what kind of challenges are also deal so like I said it's not the one-size-fits-all answer and certainly for smaller project if it's just a front end talking to a database and just a single team a few people working on it it doesn't make much sense to break it up into little pieces you're not gonna get a lot of benefit out of it from scaling or from delivery when you do decide to break up a large project there comes some costs with it you know that hopefully it's made up by the benefits of the architecture those costs are added complexity because now you've got a bunch of moving parts you've got to maintain these API contracts between the parts and so that's one of the if you want to really take advantage of micro services you need to have this strict API protocol so that pieces can Rev independently of the others you don't say my components revving so you need to go change your API you change yourself and and keep the API constant so that's one aspect you mentioned network and network latency in on network hops that starts to become another factor if it's a high throughput application that needs low latency then you need to take that into account as you design and place your micro services put the ones that are talking frequently within another close to each other monitoring becomes much more important when you're monitoring this complex distributed system and you have something that fails in one place could be the reserve and the root cause is someplace else and you need to use a lot of correlation to figure out exactly what happened after the fact so that said the like I said in many cases the bent those kinds of costs that come with it are outweighed by the benefits you get one more once again it's not a new thing cloud need you so how does sumac tow services fit into the cloud native landscape well like I said Club microservice has kind of evolved couldn't coherent you know consistently with the rise of cloud native architecture in underneath well the width so it's on demand instant scale I think if you take a look at the server's of the IT client-server era you'd say oh I want to scale out so you go to you know ITV and even for virtualization it still took weeks or months in the cloud you want to scale out and it's seconds or a minute right to scale out and add another resource and so that's what cloud native is an architecture that just intrinsically takes advantage of that kind of capability of the infrastructure and I think it has is it's less about technology so so change the culture in Libyan companies and broken down the silos you're making a lot of decisions yourself you don't have to worry about a networking team it's team and database team make everything coming back to the point of micro services in the cloud native word yeah well so like I said micro services fit nicely with that cloud native infrastructure underneath where you see the next step of micro services where things went was the rise of containerization with docker a few years ago where people said hey now can I've got this independent building block I want to package that up and and have a way to resource isolate it so that it works the same in my laptop as it does in a production environment and it can sit alongside other micro services on the same virtual machine or server and yet not be interfered with and so I've got higher reliability and reproducibility of like what I experienced in my test environment so containerization fit nicely with micro services at the same time and so that's another aspect of the evolution of that architecture and since your CTO is oh so I'll be playing a lot of buzz words so the buzz word is server less configured we're going in the funny thing is that there's a server service there if yeah yes no no you know serving you cannot have a server so can I talk about you know the the evolution of several layers and VMC know in data center services the same evolution server is something you just you know just popped it though can you talk about that yeah that's a great question because I think you know we saw microserver as a micro services and containerization with micro services the next step is service and service was really defined by the first steps of this function as a service as your functions being one example of that where it's event-driven developer provides that piece of code and the platform takes care of everything else launching the code in response to an event and managing its life cycle and the infrastructure underneath it it's called server list not because there's not servers because it does run on a server but because the developer doesn't worry about the server's and so the developer just worries about their piece of code and from a cost perspective they pay only for the piece of code as it at the time it runs in the amount of resource it consumes while it runs they don't pay for a server sitting there indefinitely they they don't have to worry about the launch time of the server and the shutdown time of the server in that cost of it's just my code gets activated it runs I just pay for that and what that provide what that enables is the developer just focuses completely on the business logic of the application and nothing about the infrastructure really and so service also represents that this move up into providing this platform where developers can just focus on the logic of their application and say here's the logic of application here now infrastructure you just go make it work you take care of running it you take care of scaling it you take care of keeping it healthy and that's first step of functions of service is leading into other types of service that are richer and richer and may be targeted at certain business scenarios so logic apps is an example of a service platform we have in Azure where a business developer not even a professional developer can write an application in a serverless way where they just write basically rules that are like Excel regular expressions and create a workflow that is powered underneath with you know servers and code that runs on the surface but they pay only for the workflow when it executes they don't worry they don't know anything about the server is it automatically scales and it's kept reliable by the underlying infrastructure and you can the huge value proposition of this of course is not just I worry about the business problem but the time to value is much shorter I'm not creating applications and deploying them um here's my logic I give it to the platform and it's take care of everything else that's because I think a lot of resources get invested and just keep the Machine well yeah instead of you know you actually focusing on what your business needs like I think 70% of resources can be extended you know just to keep just just the background here and then a cloud world the the scale we've got with the multi tenant so we have the cloud provider can absorb that kind of extra overhead but and make it very efficient very cost effective one can see that how tied are these functional service to a platform whether it's like Azure or the Google of villager or them because nowadays we talk about multi cloud we talk about interoperability so yeah great question in the case of a sure we're very focused on making sure that we address hybrid specifically and then also try to make sure customers that want to be multi-cloud can can be successful multi-cloud using our technologies so in the case of Azure functions it's actually open source and so and not only that but we make it available in Azure stack so you can take it on premises you can run it on your own you know as your functions you can deploy your own devices you can deploy it on an AWS VM if you want to you can deploy it on an edge device if you want to so that was a very conscious decision on our part to open-source the runtime and to make it possible for it to run in all these different places but what was the one of the reason behind deciding you know to keep it open so very even your competitors can use it I think if we go back to the you know the question you asked earlier about what's the value of open why does a company like Microsoft it goes back to that it goes to and in this case especially with developer runtimes and frameworks open source means that developers themselves can participate in allowing the technology also it's great as a debugging help because if there's an issue they can look right at the code if they've got you know maybe the documentation is vague on something or they misinterpret something with the code right there they can go right and look and see here's the absolute answer there's it's in black and white right in the code and in the case of of you know concerns of vendor lock and of course making an open source makes it so that somebody can say hey I can take this I can run it someplace else if I want to it's not this locked in black box that now I'm beholden forever to whoever produced it thanks for watching listen to more episodes of the news techmakers at the new stack dot io / podcasts please rate and review us on iTunes like us on YouTube and follow us on sampler see you next time [Music]