Class Service Architecture

Preface

This article was written after reading some open-source frameworks and some architectural ideas recently, just representing my own understanding and sharing, and also providing some of the same technology-making friends to see how the technology changes, and why it is a kind of service. Because I do not dare to guarantee that my thoughts must be complete service afraid of the door to axe , so I have to call the platform technology history changes

We all know, in fact, software development is nothing more than two large direction, one is the application C-s client model (need to download the app and other installation package), and the other is b-s (based on the browser access model), in fact, we have been biased in the past, that B-s browser model is not a client-side model of high technical content, In fact, in my opinion, whether it is b-s or c-s, this distinction is limited to the front-end display of the distinction (so everyone's prejudice is actually the front-end to show the technology bias), but a platform the most always want is the provision of back-office services.
What is back-office service? You use Alipay when you have a balance on the app above, the same as your login Alipay Web Access also has a balance display, and this balance is usually from the same business processing returned, so it can be said that the Web front and the app are used by the service and then provide a display effect. so our model can actually show this:

Of course, this is only one of the modes of service, the real platform is the accumulation of thousands of services, and services are not just to provide data, there is data flow to the service, such as platform orders, etc.
Now that's the pattern, what was it like before? previous software development model:

Look out and the previous pattern comparison, this development mode is directly access to the database, in the process is indeed simplified, but also encountered a problem, that is, if the project team to a new project, and this new project has a bug, this bug caused the database has a large amount of dirty data generated, Even in the presence of data for authorization (such as the absence of normal logic data, such as the next single data is not paid to write directly to the database) in order to ensure the security of the platform, has been the reliability of the platform has generated the concept of service governance, what is the service governance it.

This is just a normal mobile app development, but because the service is too much to lead to a special mix of research and development process, and if there is a conflict between services and services, such as you have to pay the payment service before the payment, you can use the order service, so in this case will be very messy, It may even lead to unpaid, and the result is that the service governance pattern is not allowed on the strong platform, which is the case:

As we can see, we have integrated the whole service into a whole, through a so-called service governance framework to integrate, and then provide services (such as the use of Alibaba Dubbo, etc.), so that developers can according to their own needs of the scene to deal with the specified business, The partition of the module can make the platform management more perfect and avoid the conflict that serves the service directly.
The picture also has a middleware for security, which is itself a service cloud, and we can use the service governance framework by dividing the service cloud and getting the same types of services together .

so in fact, like Taobao this large platform, the real core of strong support business is actually a powerful service cloud

service cloud and Message Queuing

In fact, the message queue I should focus on, but because this article is for the popularity, with the black box effect, so inconvenient to say too detailed, so I have to roughly summarize, we learned from the above services and services are directly in fact a process, the process will inevitably have communication, Message Queuing is not just used for decoupling, Sometimes also used to inform the process, of course, here we do not detail the purpose of the service cloud inside the message queue, we say that the service cloud and service Cloud Direct message Queue as shown:

In fact, you can understand that, service Cloud 1 when you want to notify the service Cloud 2 to do a certain business, will send a message to the message queue, as if a letter notification Service Cloud 2 to carry out a business, through Message Queuing can consolidate the service cloud, let the service cloud directly is a whole external service

So what is Message Queuing? In fact, the message queue can be likened to a postman, the message is called a letter, then the postman's goal is to pass this letter to the target simple description of the integration of Big data platform

What I'm specifying here is the streaming framework for storm, and of course, mapreduce and Spark, and so on, but depending on the end of the situation

This article is oriented to the public itself is not intended to write a special detailed, just hope to give you a big platform of the architecture concept, of course, people are flexible technology is dead, different combinations of methods can use different effects, if you have questions can give me a message