SaaS Series Introduction 13: SaaS System Architecture

1 System Architecture Design

System architecture in software development determines a system stability, robustness, scalability, compatibility and usability, it is the soul of the system. Architecture is at the core of the architect's focus. Good architecture is the beginning of a successful system, otherwise, good code and design will not help.

2 Introduction to the current. NET Main development Framework

L Castle

Castle is an open source project for the. NET platform, from the data Access framework ORM to the IOC container, to the MVC framework of the web layer, to AOP, which basically includes everything in the entire development process, providing a great service for us to quickly build enterprise-class applications. One of the key technologies is activerecord,facilities,monorail and so on.

Advantages: It embodies ORM, IOC, Activerecorder thought, MVC framework.

Insufficient: The framework hierarchy is not clearly divided.

L PetShop

PetShop is Microsoft's ability to use it to demonstrate the development of. NET Enterprise systems. PetShop4.0, this instance is published by Microsoft for SQL Server 2005 and Visual Studio 2005. Some of the new techniques have been used. Cached data is synchronized with database updates, new Web controls, and master applications, asynchronous communications, Message Queuing. These are very practical techniques. In PetShop, the abstract factory model was used extensively, and because of the use of master pages,membership and profile, the coding amount of the presentation layer was reduced by 25%, and the data layer was reduced by 36%.

　　

Figure 1 Architecture of PetShop4.0

PetShop4.0 in the Data Access Layer (DAL), the Dal interface abstracts out the data Access logic and uses the Dal factory as the factory module for the data Access Layer object. For DAL interface, there are SQL Server dal that support ms-sql, and Oracle Dal for Oracle. The model module contains the data entity object. It can be seen that in the data access layer, the concept of "interface-oriented programming" is fully adopted. The abstract Idal module, which is separated from the dependence of the specific database, makes the whole data access layer beneficial to the database migration. The Dalfactory module specifically manages the creation of Dal objects for easy access by the business logic layer. Both the Sqlserverdal and Oracledal modules implement the interface of the Idal module, which contains the select,insert,update and delete operations on the database. Because of the different types of databases, the operation of the database differs, and the code differs.

In addition, the abstract Idal module, in addition to the downward dependency, for its business logic layer, there is only weak dependency.

Advantages: embodies the Factory mode Orm,ioc thought,. NET Enterprise-level development.

Insufficient: No ORM thought.

L Nhibernate

Hibernate is the most extensive open source object-relational mapping framework currently in use, and it has a very lightweight object encapsulation of Java JDBC (similar to ADO), allowing programmers to manipulate the database at will using object programming thinking. Currently in the domestic Java development Community has been quite popular. And NHibernate, like Nunit,nant, is based on. NET's hibernate implementation. It mainly embodies the idea of ORM and solves the problem of persistence layer in layered development, which is very important in N-layer development.

Advantages: It embodies the ORM, the persistence layer.

Insufficient: Complex configuration, excessive reliance on XML files.

Summary of the technology used:

OR mapping ideas, layered architecture ideas, Castle-activerecorder,atlas, reflections, Design patterns (singleton mode, simple Factory mode, strategy mode), XML,IOC, framework.

3 Current Java EE main Development Framework Introduction

L Struts Frame

The struts framework is an open source product that develops Web applications based on model-view-controller (MVC) design paradigms. It uses and extends the Java Servlet API, originally created by Craig McClanahan. In May 2000, it was donated to the Apache Foundation. The Struts framework showcases a powerful custom tag library, tiled display, form inspection and i18n (internationalization). In addition, struts supports many of the description tiers, including JSP,XML/XSLT, which allow Java programmers to manipulate databases, javaserverfaces (JSF) and velocity with arbitrary object programming thinking, and also support some model layers, Includes JavaBeans and EJBS.

Here are the core elements of struts:

JSP (TAGLIB) ――>actionform――>action――> event――>ejbaction――>ejb――>dao――>database

JSP (TagLib) (forward) <――action<――eventresponse<――

Advantage: Based on MVC pattern, the structure is very good, based on JSP.

Insufficient: The scale is not very good, the logic complex large project does not apply, the frame hierarchy division is not very clear.

L Spring Framework

The spring framework is a layered JAVA/J2EE application framework based on the expert one-on-one Java EE Design and distribution code. The spring Framework provides a simple development technique for automating the processing of a large number of attribute files and helper classes in a project.

Spring is an open-source framework that was created by Rod Johnson and described in his book, "The Java EE Design Development Programming Guide". It is created to address the complexities of enterprise application development. Spring makes it possible to use basic javabeans to accomplish things that were previously only possible by EJBS. However, the use of spring is not limited to server-side development. From the standpoint of simplicity, testability, and loose coupling, any Java application can benefit from spring.

The main features of the spring framework include:

1 powerful JavaBeans-based configuration management, using the Inversion-of-control (IoC) principle.

21 Core bean factories that can be used in any environment, from applets to the Java EE container program.

The 3 General abstraction layer is suitable for database transaction management, allows pluggable transaction managers, and easily divides the boundaries of each transaction without having to deal with low-level problems.

4 A very meaningful JDBC Abstraction layer for exception handling.

5 integrated with Hibernate, DAO implements support and transaction policies.

Advantages: It embodies the idea of Java EE, container, lightweight, inversion of control and face-facing.

Insufficient: complex structure, difficult to understand.

L Hibernate Frame

Hibernate is an open-source object-relational mapping (ORM) framework that provides a very lightweight object encapsulation of JDBC, providing an easy-to-use framework for mapping an object-oriented domain model to a traditional relational database. So that Java programmers can use object programming thinking to manipulate the database at will. It is not only responsible for mapping from Java classes to database tables (and SQL data types from Java data types), but also provides data query and retrieval capabilities and greatly reduces development time spent on SQL and JDBC manual data processing. Most revolutionary of all, hibernate can be used to replace CMP in the Java EE architecture of EJB, and accomplish the task of data persistence.

Hibernate's goal is to alleviate the developer's programming tasks associated with a large amount of common data persistence. Hibernate is also able to adapt to the development process, whether it is just beginning to design or from a ready-made database. Hibernate automatically generates SQL, freeing developers from the tedious task of manually processing result sets and object conversions, and porting applications to all SQL databases. It also provides transparent persistence, and the only requirement for persistent classes is to implement a parameterless constructor.

Advantages: The main application in the EJB layer, the configuration is strong, flexible, simplifying the database operation.

Insufficient: Difficult to configure.

4 Common software Architectures

L Three-tier architecture

In the design of software architecture, layered structure is the most common and most important structure. The layered structure is generally divided into three layers, from bottom to top: The data access layer, the business logic layer (or become the domain layer), the presentation layer:

　　

Figure 23 Layer Architecture

Data Access layer: sometimes also known as the persistence layer, its function is mainly responsible for database access. The simple argument is to implement the select,insert,update,delete of the data table. If you want to add an ORM element, it will include the mapping between the object and the data table, and the persistence of the object entity.

Business Logic Layer (BusinessRules): is the core of the entire system, which is related to the business (domain) of this system. In the case of STS system, the related design of business logic layer is related to the logic of sales tracking. Structurally, it encapsulates the related operations of the data access layer. This layer consists primarily of classes that implement specific business logic.

Presentation Layer (WebUI): Is the UI part of the system that is responsible for user interaction with the entire system. In this layer, the ideal state is that the business logic of the system should not be included. Represents the logical code in a layer, which is related only to interface elements. In the current project, it is designed with ASP., so it contains many Web controls and related logic.

L Five-tier architecture

SaaS software architecture can also be divided into five tiers, from top to bottom: User interface layer (presentation layer), Business logic layer, universal layer, Application framework layer, remote access (WebService) layer, data access layer:

　　

Figure 3 is based on Microsoft. NET Architecture Design

User interface Layer (UI)

The user interface layer is the user interface that is directly manipulated. This layer is composed of the interface appearance, form controls, frames, and other parts. The user interface layer is responsible for the interaction of the consumer with the entire system. In this layer, the ideal state is that the business logic of the system should not be included. Represents the logical code in a layer, which is related only to interface elements. In the current project, it is designed with ASP., so it contains many Web controls and related logic.

² interface appearance including Skip (skin), Images (picture), CSS (style sheet)

² form controls mainly include common forms, user-defined controls.

² The framework mainly includes the master page, frame page.

² Other main include JavaScript files, DLL files, report reports, schema database, model development template.

Business Logic Layer (BusinessRules)

Is the core of the entire system, and it is related to the business (domain) of the system. In the case of STS system, the related design of business logic layer is related to the logic of sales tracking. Structurally, it encapsulates the related operations of the data access layer. This layer consists primarily of classes that implement specific business logic.

²blfactory Business Logic Factory

²IBL Business Logic Interface

²businessrules Business Logic Implementation

Universal Layer

The general-purpose layer runs through the presentation and business logic layers of the entire project. The main storage of the project is the more general definition of constants and common services (service), here refers to the service is the current project business logic common method, we write them in the corresponding static class. Provided in the form of a service.

Commonlayer: Storage of common constants and methods.

Data Access Layer

The layer structure is the most complex, consisting mainly of the following layers: The data Access Factory layer (dalfactory), the Data Access Interface layer (Idal), the custom query layer (PERSISTENCEFACADE), the staging layer (Dataaccesslayer), the Data Persistence layer ( Persistencelayer).

The following is introduced from the bottom up:

²persistencelayer layer, which is the lowest level of the frame design (except the application framework layer). It is primarily responsible for using ORM ideas to target physical databases. The simple thing is to map the database table to an entity class and map the corresponding fields to the properties of the class. In this way the physical database is completely transparent to the developer, and the idea of applying ORM is completely free of the physical database. and is independent of the database specific implementation.

² specifically implement our lightweight data access component Activerecorder implemented under the well-known open source project Castle.

²persistencefacade layers and Idal, which define all the query methods used in the project. Corresponds to the data entity defined by the Persistencelayer layer. In these word-defined query classes, you can apply any combination of the three query methods provided by Activerrecorder (simple interfaces provided by Activerrecorderbase, simple query SimpleQuery, custom query customerquery). And each of the classes here implements the interface defined by the Idal interface layer.

²dalfactory layer, as the data access factory, through. The reflection mechanism of net calls the related operation in the data access component composed of Idal and Persistencefacade.

²dataaccesslayer temporary layer. It is completely unnecessary to declare this layer first. Because we can not write any SQL statements in the project. All SQL is replaced with HQL. This layer is designed to allow for the technical transition of people in the project team. This layer can operate the database through SQL (not recommended). This layer will no longer be available when the architecture is stable.

Application framework Layer (framework)

The purpose of this layer is technical precipitation. Moving things that are common between projects into the application framework layer to achieve code reuse. This layer can be black-boxed later. can include generic components.

²framework: Accumulate some methods and controls that can be abstracted

²MSMQMESSAG: Implementation of the message processing queue

²pager: Universal Flip Class

²report: General report Class

²controls: Control processing class

²dataformat: Data Format conversion class

²webui: Page Processing class

²validate: Data Check

²object: Conversion and access between objects

5 Benefits of Tiered architecture

1, the developer can only focus on the entire structure of one of the layers;

2, can easily use the new implementation to change the original level of implementation;

3. Can reduce the dependence between layer and layer;

4, in favor of standardization;

5, facilitate the reuse of the logic of each layer.

In summary, layered design can achieve the following objectives: Decentralized attention, loose coupling, logical multiplexing, and standard definitions.

A good layered structure can make the developer's Division of labor more explicit. Once the interfaces between the various levels are defined, developers who are responsible for different logic designs can distract and go hand in hand. For example, UI staff only need to consider the user interface experience and operation, the domain designer can focus only on the design of business logic, and database designers do not have to be cumbersome user interaction and headache. Each developer's task is confirmed and development progress can be improved quickly.

The benefits of loose coupling are obvious. If a system does not have a hierarchy, then the respective logic is tightly intertwined, interdependent with each other, who can not be replaced. Once the change occurs, the reaching, the impact on the project is extremely serious. Reducing the dependency between layers and layers can not only guarantee the future scalability, but also have obvious advantages in reusability. Once each function module has a unified interface defined, it can be called by each module without having to develop the same functionality repeatedly.

The standard is also essential for the design of a good layered structure. Only on the basis of a certain degree of standardization, the system is extensible, replaceable. The communication between layers and layers must also ensure the standardization of interfaces.

"Gold is no can't pure, nobody is perfect", the layered structure also inevitably has some defects:

1, reduce the performance of the system. This is self-evident. Without a tiered structure, many businesses can access the database directly to get the data, and now they have to do it through the middle tier.

2. Sometimes cascade changes are caused. This kind of modification is especially reflected in the top-down direction. If you need to add a feature in the presentation layer to ensure that the design conforms to the layered structure, you may need to add code to the appropriate business logic layer and the data access layer.

6 Software Architecture View

Philippe Kruchten wrote in his book, "An introduction to the rational unification process":

An architectural view is a simplified description of a system that is seen from a certain perspective or point, and describes a particular aspect of the system and omits entities unrelated to this aspect.

In other words, the architecture to cover a lot of content and decision-making, beyond the human brain "overnight" capabilities, so the "divide and conquer" approach from different perspectives to design, but also for the understanding of software architecture, communication and archiving to provide convenience.

　　

Figure 4 4+1 View method presented by Philippe Kruchten

The different architectural views of this approach carry different architectural design decisions, supporting different goals and uses:

L Logical View: When adopting an object-oriented design approach, the logical view is the object model.

L Development View: Describes the static organization of software in the development environment.

L Processing Views: Describes the concurrency and synchronization aspects of the system design.

Physical view: Describes how software is mapped to hardware and reflects the design of the system in terms of distribution.

　　

Figure 5 Architecture design for different requirements using the 4+1 view method

Logical view. The logical view focuses on features that include not only user-visible features, but also "accessibility modules" that must be provided to implement user functionality, such as logical layers, functional modules, and so on.

Development view. The development view focuses on packages, including not only the source programs to be written, but also the third-party SDKs and out-of-the-box frameworks that can be used directly, the class library, and the system software or middleware that the developed system will run on. There may be some mapping between the development view and the logical view: for example, the logical layer typically maps to multiple packages, and so on.

Process the view. Processing views focus on the runtime concepts of processes, threads, objects, and related concurrency, synchronization, communication, and so on. Dealing with the relationship between the view and the development view: The development view typically focuses on the static dependencies of the package at compile time, and these programs behave as objects, threads, processes, and the interaction of these runtime units is the focus of the processing view.

Physical view. The physical view focuses on how the target program and its dependent runtime and system software are ultimately installed or deployed to the physical machine, and how the machines and networks are deployed to meet the reliability and scalability requirements of the software system. Relationship between physical and processing views: processing views pay particular attention to the dynamic execution of the target program, while the physical view attaches importance to the static location of the target program; The physical view is an architectural view that takes into account the interaction between the software system and the entire IT system.

7 Summary

This paper introduces the architecture design method of SaaS. Through the introduction of the main development framework of. NET and the introduction of the main development framework of Java EE, we can analyze the advantages and disadvantages of each.

At the same time, the hierarchical model of software architecture is introduced, and the core value of software architecture is embodied by concrete layering, and the model of architecture can be applied in our development.

SaaS Series Introduction 13: SaaS System Architecture