Essential Software Architecture - Chapter 1 Understanding Software Architecture - Summary

Essential Software Architecture - Chapter 1 Understanding Software Architecture - Summary

• 1.
  Essential Software Architecture Chapter 1: Understanding Software Architecture Summary 1.1 What is software Architecture ● During the last fifteen years we have seen how software architecture is topic of relevant importance. ● IASA (International Association of Software Architects) is an association for Software Architects. ● Architecture is an overused and least understood term in professional software development circles. ● A term is gradually becoming vacuous when it becomes part of the vernacular of the software industry sales force. 1.2 Definitions of Software Architecture ● Define a term like Software Architecture is always a potentially dangerous activity. ○ There really is no widely accepted definition by the industry. ● Definitions of Software Architecture: ○ IEEE: Architecture is defined by the recommended practice as the fundamental organization of a system, embodied in its components, their relationships to each other and the environment, and the principles governing its design and evolution. ■ Architecture captures system structure in terms of components and how they interact. ■ It also define system-wide design rules and considers how a system may change. ○ Software Architecture in Practice (2nd Edition)1: ■ The software architecture of a program or computing system is the structure or structures of the system, which comprise software elements, the externally visible properties of those elements, and the 1 [L.Bass, P.Clements, R.Kazman, Software Architecture in Practice (2nd edition), Addison-Wesley 2003]
• 2.
  relationships among them. ○ From Garlan and Shaw’s early influential work: ■ Software Architecture goes beyond the algorithms and data structures of the computation; designing and specifying the overall system structure emerges as a new kind of problem. Structural issues include gross organization and blog control structure; protocols for communication, synchronization, and data access; assignment of functionality to design elements; physical distribution; composition of design elements; scaling performance; and selection among design alternatives. 1.2.1 Architecture Defines Structure ● Much of an architect’s time is concerned with how to sensibly partition and application into a set of interrelated components, modules, objects or whatever unit of software partitioning works for you. ● An architecture must be designed to meet the specific requirements and constraints of the application it is intended for. ● For example: ○ A requirement for an information management system may be that the application is distributed across multiple sites, and a constraint is that certain functionality and data must reside at each site. ○ An application’s functionality must be accessible from a web browser. ● All these impose some structural constraints: ○ Site-specific, ○ Web server hosted ● ...and simultaneously open up avenues for considerable design creativity in partitioning functionality across a collection of related components. ● In partitioning an application, the architect assigns responsibilities (the tasks a component can be relied upon to perform within the application) to each constituent component. ● All the chunks, then, must be ensamble to provide the required functionality. ● Responsibility-driven design2 is a technique from object-orientation that can be used effectively to help define the key components in an architecture. ○ This approach provides the tools and techniques that emphasize behavioral modeling using objects, responsibilities and collaborations. ● A key structural issue for nearly all applications is minimizing dependencies between components, creating a loosely coupled architecture from a set of highly cohesive components. ● A dependency exists between components when a change in one potentially forces a change in others. By eliminating unnecessary dependencies, changes are localized 2 http://www.wirfs-brock.com/Design.html
• 3.
  and do notpropagate throughout an architecture. ● Excessive dependencies are simple a bad thing. ○ They make it difficult to make changes to systems, ○ More expensive to test changes, ○ they increase build times, and ○ They make concurrent, team-based development harder. 1.2.2 Architecture Specifies Component Communication ● When an application is divided into a set of components, it becomes necessary to think about how these components communicate data and control information. ● Components can communicate through: ○ Straightforward method calls ○ Implicit method innvocation ○ Remote method invocation, and so on. ● They may execute in different threads or processes, and communicate through synchronization mechanisms. ● Or multiple components may need to be simultaneously informed when an event occurs in the application’s environment.
• 4.
  ● Architectural patternsand styles3 has catalogued a number of successfully used structures that facilitate certain kinds of component communication. ● Patterns provides a well-known and accepted reusable architectural blueprints that describe the structure and interaction between collections of participating components. ● Each pattern has characteristics that make it appropriate to use in satisfying particular types of requirements. For example: ○ The client-server pattern has several useful characteristics, such as synchronous request-reply communications from client to server, and servers supporting one or more clients through a published interface. ○ Client-server architectures must also provide a mechanism for clients to locate servers, handle errors, and optionally provide security on server access. ● The power of architecture pattern stems from their utility, and ability to convey design information. ● Large systems tend to use multiple patterns, combined in ways that satisfy the architecture requirements. ● When an architecture is based around patterns, it also becomes easy for team members to: ○ understand a design, ○ as the patterns infers component, ○ communications, and ○ abstract mechanisms that must be provided 1.3 Architecture Addresses Nonfunctional Requirements ● Nonfunctional requirements are the ones that don’t appear in use cases. ● Rather than define what the application does, they are concerned with how the application provides the required functionality. ● There are three distinct areas of nonfuncional requirements: ○ Technical constraints: ■ They constrain design options by specifying certain technologies that the application must use: ● “We have only Java developers, so we must develop in Java” ● “The existing database runs on Windows XP only” ■ These are usually nonnegotiable. ○ Business constraints: ■ These too constraint design options, but for business, not technical reasons. e.g.: 3 Patterns and styles are essentially the same thing.
• 5.
  ● “In orderto widen our potential customer base, we must interface with XYZ tools”. ● “The supplier of our middleware has raised prices prohibitively, so we’re moving to an open source version”. ■ Most of the time, these too are nonnegotiable. ○ Quality attributes: ■ These define an application’s requirements in terms of: ● Scalability, ● Availability, ● Ease of change, ● Portability, ● Usability, ● Performance, and so on. ● An application architecture must therefore explicitly address these aspects of the design. ● Architects need to understand the functional requirements, and create a platform that supports these and simultaneously satisfies the nonfunctional requirements. 1.3.1 Architecture Is an Abstraction ● A marketecture4 is an excellent vehicle for facilitating discussion by stakeholders during design, build, review, and of course the sales process. ● It’s easy to understand and explain and serves as a starting point for deeper analysis. ● A thoughtfully crafted marketecture is particularly useful because it is an abstract description of the system. ● In reality, any architectural description must employ abstraction in order to be understandable by team members and project stakeholders. ○ This means that unnecessary details are suppressed or ignored in order to focus attention and analysis on the salient architectural issues. ○ This is typically done by describing the components in the architecture as black boxes, specifying only their externally visible properties. ● One of the most powerful mechanisms for describing and architecture is hierarchical decomposition. ○ Components that appear in one level of description are decomposed in more detail in accompanying design documentation. ● Fig. 1.2 depicts a very simple two-level hierarchy using an informal notation, with two of the components in the top-level diagram decomposed further. 4 http://en.wikipedia.org/wiki/Marchitecture
• 6.
  ● Different levelsof description in the hierarchy tend to be of interest to different developers in a project. ● The architecture clearly partitions the responsibilities of each team, defining the dependencies between them. ● In this hypothetical example, the architect has refined the design of two of the components, presumably because some nonfunctional requirements dictate that further definition is necessary. ● Perhaps an existing security service must be used, or the Broker must provide a specific message routing function requiring a directory service that has a known level of request throughput. ● In the above example, the Client component is not partitioned, because the internal structure and behavior of the client is not significant in achieving the application’s overall nonfunctional requirements. ○ The Client component could possibly be the most complex in the application. It might have an internal architecture defined by its design team, which meets specific goals for the Client component. ● It’s not necessary for the architect to complicate the application architecture with such issues, as they can be safely left to the Client design team to resolve. 1.3.2 Architecture Views ● A software architecture represents a complex design artifact. ● The term “architecture views” is a way of describing and understanding an architecture based on the following four views: ○ Logical View:
• 7.
  ■ This describesthe architecturally significant elements of the architecture and the relationship between them. ■ The logical view essentially captures the structure of the application using class diagrams or equivalents. ○ Process view: ■ This focuses on describing the concurrency and communications elements of an architecture. ■ In IT applications, the main concerns are describing multithreaded or replicated components, and the synchronous or asynchronous communication mechanisms used. ○ Physical view: ■ This depicts how the major processes and components are mapped on to applications hardware. ■ It might show, for example, how the database and web servers for an application are distributed across a number of server machines. ○ Development view: ■ This captures the internal organization of the software components, typically as they are held in a development environment or configuration management tool. ■ For example, the depiction of a nested package class hierarchy for a Java application would represent the development view of an architecture. ● These views are tied together by the architecturally significant use cases: often called scenarios. ● By working through the steps in a particular use case, the architecture can be “tested”, by explaining how the design elements in the architecture respond to the behaviour required in the use case. ● Krutche’s paper and other works, recommend capturing an architecture model using three different views [alternative approach]: ○ Module: ■ This is a structural view of the architecture, comprising the code modules such as classes, packages, and subsystems in the design. ■ It also captures: ● module decomposition, ● inheritance, ● associations, and ● aggregations. ○ Component and connector: ■ This view describes the behavioral aspects of the architecture. ■ Components are typically: ● Objects, ● threads or processes. ■ Connectors describe how the components interact.
• 8.
  ● Common connectorsare sockets, middleware like CORBA or shared memory. ○ Allocation: ■ This view shows how the processes in the architecture are mapped to hardware, and how they communicate using networks and/or databases. ■ It also captures a view of the source code in the configuration managements systems, and who in the development group has responsibility for each modules. ● The terminology used in “Views and Beyond” is strongly influenced by the architecture description language (ADL) research community. 1.4 What Does a Software Architect Do? ● The environment that a software architect works in tends to define their exact roles and responsibilities. ● FOUR (4) skills for a software architect, regardless of their professional environment: ○ Liaison: ■ Architects play many liaison roles. ■ They liaise with: ● Customers ● Clients ● Technical team ● Business and requirements analysts. ■ They justify designs, decisions and cost with the management deparment. ■ They liaise with sales, to help promote a system to potential purchasers or investors. ■ Software architect must a simple terminology depending the stakeholder. ○ Software Engineering: ■ Deep understanding of the technology domains (relevant to context of applications they work on). ■ Know how to exploit new technologies and features into the working application. ■ Just as importantly, good architects know what they don’t know, and ask other with greater expertise when they need information. ○ Risk Management: ■ Good architects tend to be cautious. ■ They are constantly enumerating and evaluating the risks associated with the design and technology choices they make. ■ They document manage these risks in conjunction with project sponsors and management.
• 9.
  ■ They try to make sure no unexpected disaster occur. ● Architects play a central role in software development, and must be multiskilled in software engineering, technology, management and communication. 1.5 Architects and Technologies ● Architect must make design decisions early in a project lifecycle. ● That decisions are so difficult. ● Due to the difficulty of validating early design decisions, architects sensibly rely on tried and tested approaches for solving certain classes of problems. ● Through architectural patterns architects can reduce risk leveraging successful designs with known engineering attributes. ● Patterns are an abstract representation of an architecture. They can be realized in multiple concrete forms. ● Patterns only describe solutions in abstract terms. ● Frameworks are implementation of design patterns. They come to rescue. ● As Fig. 1.3 depicts, several classes of COTS technologies are used in practice to provide packaged implementations of architectural patterns for use in IT systems. ● Different COTS technology implementations have different sets of strengths and weakness and costs, and consequently will be better suited to some types of applications than others. ● The difficulty for architects is in understanding these strengths and weakness early in the development cycle for a project. ○ ...and choosing an appropriate reification of the architectural patterns they need. ● The history of the software industry is littered with poor choices and subsequent failes projects.
• 10.
  ● Eoin Woods says: ○ Software architecture is the set of design decisions which, if made incorrectly, may cause your project to be cancelled. 1.6 Architect Title Soup ● Scan job advertisements: ○ Chief Architect: ■ Typically a senior position who manages a team of architects within an organization. ■ Operates at: ● a broad ● organizational level ● and coordinates efforts across system applications, and product lines. ■ Very experienced. ■ Rare combination of deep technical and business knowledge. ○ Product/Technical/Solution Architect: ■ Have a deep knowledge of how some important piece of software really works. ○ Enterprise Architect: ■ More business-focus role. ■ Use various business methods and tools to understand, document, and plan the structure of the major systems in an enterprise. 1.8 Further Reading 1.8.1 General Architecture
• 11.
  5 1.8.2 Architecture Requirements I.Jacobson, M. Christerson, P. Jonsson, G. Overgaard. Object-Oriented Software Engineering: A Use Case Driven Approach. Addison-Wesley, 1992. R. Wirfs-Brock, A. McKean. Object Design: Roles, Responsibilities, and Collaborations. Addison-Wesley, 2002. 1.8.3 Architecture Patterns F. Buschmann, R. Meunier, H. Rohnert, P. Sommerlad, M. Stal,. Pattern-Oriented Software Architecture, Volume 1: A System of Patterns. John Wiley & Sons, 1996. D. Schmidt, M. Stal, H. Rohnert, F. Buschmann. Pattern-Oriented Software Architecture, Volume 2, Patterns for Concurrent and Networked Objects. John Wiley & Sons, 2000. 5 From Essential Software Architecture by Ian Gorton
• 12.
  Two recent booksthat focus more on patterns for enterprise systems, especially enterprise application integrations, are well worth a read. M. Fowler. Patterns of Enterprise Application Architecture. Addison-Wesley, 2002. G. Hohpe, B. Woolf. Enterprise Integration Patterns: Designing, Building, and Deploying Messaging Solutions. Addison-Wesley, 2003. 1.8.4 Technology Comparisons P. Tran, J. Gosper, I. Gorton. Evaluating the Sustained Performance of COTS based Messaging Systems. in Software Testing, Verification and Reliability, vol 13, pp 229–240, Wiley and Sons, 2003. I. Gorton, A. Liu. Performance Evaluation of Alternative Component Architectures for Enterprise JavaBean Applications, in IEEE Internet Computing, vol.7, no. 3, pages 18–23, 2003. A. Liu, I. Gorton. Accelerating COTS Middleware Technology Acquisition: the i-MATE Process. in IEEE Software, pages 72–79,volume 20, no. 2, March/April 2003.