- The primary purpose of architecture is to support the life cycle of the system. Good architecture makes the system easy to understand, easy to develop, easy to maintain, and easy to deploy. The ultimate goal is to minimize the lifetime cost of the system and to maximize programmer productivity.
Clean Architecture: A Craftsman's Guide to Software Structure and Design by Robert C. Martin
- 1 System Overview
- 2 Java Software Structures
- 3 Quality Attributes in Software Architecture
- 4 OSEE Bundle Architecture
- 5 OSEE Bundle Structure/Layering
- 6 Open Standards Used in OSEE
- 7 OSGi Declarative Services
- 8 OSEE Components
- 9 OSEE Functional Use Cases
- 10 OSEE Deployment
- 11 Links
The Open System Engineering Environment (OSEE) is an integrated, extensible tool environment for large engineering projects. OSEE is more than an integrated development environment (IDE), but is an integrated product life-cycle development environment. The system captures project data into a common user-defined data model providing bidirectional traceability, project health reporting, status, and metrics which seamlessly combine to form a coherent, accurate view of a project in real-time. By building on top of a central data model, OSEE provides an integrated configuration management, requirements management, implementation, testing, validation, and project management system. All of the components work together to help an organization achieve lean objectives by reducing management activities, eliminating data duplication, reducing cycle-time through streamlined processes, and improving overall product quality through work flow standardization and early defect detection.
OSEE is customizable to meet the needs of the project. The teams working on the project, the roles they perform, and the processes they follow are all configurable within OSEE. Traceability is maintained from requirements through acceptance testing. The OSEE High Level System Overview diagram illustrates the range of users and the roles they may perform, which encompasses the entire product life-cycle
- Requirements Engineer can create, manage, track requirements
- Implementation Engineer may design, implement, integrate, unit test the product from software to hardware.
- Test Engineer may perform different levels of integration or acceptance testing, and provide comprehensive test reporting.
- Project Manager can track the progress of the product and gather customized metrics in order to best gain insight into the project status.
- System Administrator manages the user accounts and IT needs.
- Support Engineer can trace progress on customer issues
- Maintenance Engineer can run validation, verification, or diagnostic tests on the target environment.
Java Software Structures
Levels of code organization
- bundles contain packages
- packages contain types (classes and interfaces)
- classes contain methods
- methods contain blocks
- blocks contain statements
- statements contain expressions
- blocks contain statements
- methods contain blocks
- classes contain methods
- packages contain types (classes and interfaces)
- explicitly declare the packages they use (their external dependencies) and the packages they export for use by other bundles
- are organized hierarchically and provide a namespace for the types they contain
- define objects and how they are created
- defines the parameters it accepts and the type of value it returns (or void)
- forms a complete unit of execution
- expression statements - Assignment, method invocation, object creation
- declaration statements - type variableName = value; i.e. int daysInWeek= 7;
- control flow statements - decision making, looping, and branching, enabling conditional execution of blocks of code
- decision-making statements (if-then, if-then-else, switch)
- looping statements (for, while, do-while)
- branching statements (break, continue, return)
Responsibilities and Constraints
At various levels of the software structure (bundles, packages, types, and methods) the responsibilities and constraints of that structure is documented using Javadoc. For bundles, this information is in the package-info.java file in the top level package of the bundle. For packages see the package-info.java file in the corresponding package. For classes and methods their Javadoc is at the top of the class or method.
Quality Attributes in Software Architecture
- Availability - part of reliability and is expressed as the ratio of the available system time to the total working time
- Extensibility - The services/components provided with OSEE will be reusable and/or extendable.
- Flexibility - The tailoring of OSEE for a specific project can be accomplished through dynamic configuration.
- Interoperability - facilitates integration with third-party systems
- Maintainability - easily change to meet new business requirements; easily change to meet non-functional requirements
- Openness - OSEE uses open standards to enable quality, standardized, and well thought-through interfaces.
- Performance - low latency for individual requests
- Reliability - system's ability to continue to operate in the face of various adverse conditions
- Reusability -
- Scalability - The system will scale to handle large numbers of simultaneous events and managed projects.
- Security - prevent inappropriate access to information and other malicious actions
- Supportability - support diagnosis of issues health checking
- Testability - the system allows performing automated tests with high percentage of code coverage
- Usability - intuitive interface that is discoverable and minimizes the overall effort required by the user
OSEE Bundle Architecture
- ats.api - Models and Interfaces (ide/server)
- ats.core - Business logic, implementation of interfaces (ide/server)
- ats.core.test - Junit (does not require server and db)
- ats.ide - Eclipse / Windows user interface; Implementation of client specific interfaces
- ats.ide.demo - Demon Database business logic
- ats.ide.demo.feature - List of bundles that are needed for demo
- ats.ide.help.ui - Help User Interface - Generated from wiki
- ats.ide.integration.tests - AtsIde_Integration_TestSuite - Demo Db Init and Integration tests (requires server/db)
- ats.rest - Integration of Interfaces specific to server; REST implementations (should be renamed to ats.server)
- ats.rest.test - Tests for Server specific implementations (should be renamed to ats.server.test)
- Anything with "rest" == server
- Anything with "integration" == integration tests require server/db; run with launch configs
- Anything with ".test" (not integration) == Pure Java JUnit test
- Anything with "ide" == Eclipse bundles w/ OS User Interfaces
- Anything with "api" or "core" = shared client and server
OSEE Bundle Structure/Layering
- ATS - Configuration Management
Def: Process/Workflow control of managing data - Depends on Define
- Define - Content Management
Def: Generic capabilities/editors of managing content - Does NOT depend on ATS - Traceability - Publishing - Reporting - Generic Editors - Rendering - Templates - Word everything
- Orcs - Revision Control System including all things Data Model
- Server only - Art, Attr, Rel, Appl, Branch
- Client only - Art, Attr, Rel, Appl, Branch
- Next Steps
- framework.ui.skynet should be rolled into define.ide
Open Standards Used in OSEE
OSGi Declarative Services
OSGi is the Dynamic Module System for Java. Its services model enables application and infrastructure modules to communicate locally and distributed across the network.
OSGi Declarative Services are explained well here Getting Started with OSGi Declarative Services which is summarized below:
OSGi services use a publish-find-bind mechanism. A bundle can provide/publish a service implementation of a given interface (type) for other bundles to consume. With declarative services, services are not registered or consumed programmatically. Instead, a Service Component is declared via a Component Description in an XML file in the OSGI-INF folder. The Component Description is processed by a Service Component Runtime (SCR), e.g. Equinox DS or Felix SCR) when a bundle is activated.
- Service API
- Service Provider
- Service Consumer
OSGi services are dynamic so, the service consumer must react to life cycle events. Service Components have their own lifecycle, which is contained in the life cycle of a bundle. OSGi service retrieval is type-safe since it is done based on an interface (type)
OSGi Bundle Diagnostics Console Commands
- lb or ss - to list bundles and their states
- bundle <bundleId> -
- diag (try with no arguments)
OSGi Component Diagnostics Console Commands
- list | grep unsatisfied
- Use component.id from list command for unsatisfied components
- list <bundle.id>
- list | grep <component.name>
- info <component.id>
- inspect capability service
OSGi best practices
- use declarative services and no Activators.
- Bundle-ActivationPolicy lazy
- Don't specify a Component name. Per OSGi Compendium Release 7 126.96.36.199, if not specified, the name of a Component is the fully qualified type name of the class being annotated.
OSEE follows a client server architecture with a thin client and one to N servers. The client can exist in two different forms: One, a web browser client; Two, an Eclipse based IDE. The OSEE Server is built utilizing the Eclipse Equinox OSGi framework. All instances of the server attach to a single centralized data repository.
At the core of OSEE Application Server is the Object Revision Control System Framework. On top of the framework sits four core components: Action Tracking System (ATS), Define, Coverage, and Open System Engineering Test Environment (OTE). The User Management component, which allows for user authentication, verification and role based access control (RBAC), is used by all of the OSEE components.
- Logging - Provide a consistent logging mechanism
- Console - User ability to issue command line operations on the server
- JAX-RS - Java REST API for web services
- Event Management - ActiveMQ is used for event messaging
The OSGi Framework Linked above provides the underlying support for the OSEE Framework. The Eclipse implementation of OSGi is called Equinox. The Equinox framework supports Extensions, Services, Declarative Services and Spring-OSGi. Since OSEE is built on top of this extendable foundation, it inherits these capabilities. OSEE developers have primarily taken advantage of Declarative Services.
The layered approach provides extensibility to OSEE. The IDE Client is extended through the use of Eclipse Extension Points. The OSEE Application Server is extended through Declarative Services. As described in the following sections, the data model is abstracted one level up, so that a broad variety of specific data can be expressed in the data model. The database interface complies with standard SQL-99 ("with" clause is used so SQL-92 is not sufficient), so relational databases providing SQL-99 compliance can be substituted in through the use of JSON configuration files provided in OSEE.
Object Revision Control System
The heart of the OSEE Framework is the Object Revision Control System (ORCS). ORCS provides the foundation the rest of the components are built on top of. The key capabilities provided by ORCS are:
- Object Management - The definition and persistence of any object of any simple or complex type can be managed, and type safety is ensured on all operations.
- Data Model - the underlying data model is configurable per project
- Version management - baselines and branching for a project are managed
- Generic UI editor framework - customized editors can be created
- Transaction Management
- Data Import/Export - external systems can supply or consume data
The object management provides the core building blocks for the centralized data model. The core building block for the data model consists of:
- Artifact - The fundamental object in OSEE. All data objects stored within OSEE are artifacts. Artifacts are strongly typed and can store any data throughout the systems engineering lifecycle. Artifacts have a Description, Type, and set of 1..n Attributes. Any type of data can be stored as an artifact; not only systems engineering data (such as processes and requirements), but also anything from meeting minutes to architecture diagrams.
- Attribute - A specific piece of data attached to an Artifact. Attribute consists of Description, Type, default value (optional), Min occurrences in an artifact, Max occurrences in an artifact. An Attribute can be a basic type or a more complex type like another Artifact.
- Relation - Defines the relationship (link) between two artifacts. The relation is strongly typed, which means it can only be used to link the specified artifact types. The relation allows the multiplicity between the two types to be specified as: 1..1, 1..*, *..1, *..*
Version management allows for the parallel development of different variations of a product, as well as the sharing of common information across similar products. Changes made to one version baseline can be merged to another version baseline in order to maintain commonality as desired.
Action Tracking System (Details)
The Action Tracking System (ATS) is a tightly integrated tracking system that manages changes throughout the different aspects of a product's lifecycle. ATS provides integrated change management to all OSEE applications through customizable work processes (workflows) and ensures traceability from start to completion. ATS utilizes the core capabilities provided by the ORCS layer.
ATS is highly configurable and can be configured to meet any project's work tracking needs. The level of detail of work items, team organization, and process to complete work item types are all configurable for a project. The configuration is realized through the use of the data model to create the core building blocks of ATS: Action, Actionable Item, Team Definition, Workflow Definition, Task, and Version.
At the highest level, an item of work to be completed is referred to as an Action. Actions are created as work is needed for a project.
A project can specify a work hierarchy for the different kinds of work tasks that need to be performed and tracked. Each defined work category is referred to as an Actionable Item (AI). An Action can be composed of a single or multiple AIs.
A team can be assigned to work on an AI. The team definition is similar to an organization chart, or a logical grouping of teams that perform certain types of work.
Each Team Definition has a Workflow Definition (or state machine) that defines the process that team uses to track and complete the work. Each state of the workflow can have configured conditions or fields that are required to transition. A Review can be attached to a state and can block the transition until successfully completed.
A Task is the lowest level of work, and is used to allocate the work to individuals. A Task can be associated with a particular state or a state can have multiple tasks that need completing before the workflow can advance to the next state.
A Version is used to group a set of Actions together into a "build", "release", "edition", etc. A common set of actions can apply to more than one version enabling data sharing across similar or variant projects. The version capability relies on the version management framework.
Status data associated with tasks can be used to create metrics that roll-up to the Workflow, which can roll up to the Team, which can roll-up to the Action, which roll-up to the project. Metrics can be obtained for any specified grouping within the project (e.g. Team, Version, etc.)
Define is the requirements management component of OSEE. Define provides support for concurrent and distributed requirements development. Requirements can be imported from other sources to provide comprehensive and coherent requirements management across the product life-cycle.
- can be hierarchical
- can be integrated into the processes and workflows
- will have bidirectional traceability from beginning to end of the product life-cycle
- will have meaningful review metrics
- can support parallel project development
Coverage provides for the configuration management and tracking of coverage disposition efforts throughout a project. OSEE allows for the configuration of what is tracked for verification and validation on the project. For example, in a software project the lines of code in the software can be exercised through software tests, and tools can determine how many of the software lines of code where executed during the tests. A report is generated and can be imported by Coverage and a complete coverage report generated.
The Coverage is configurable through the creation of a Coverage Package. The Coverage Package configuration includes all the inputs (unit tests, test scripts, coverage test report imports), traceability (results to tests, tests to requirements) and outputs (reports, metrics) desired. The Coverage Package will also allow for checking the differences between test runs by comparing different instances of test result imports.
Open System Engineering Test Environment
The Open System Engineering Test Environment (OTE) is an integrated approach to product testing, and follows a client/server architecture. The OTE client is part of the OSEE client, while the OTE Server is separate from the OSEE Application Server.
The OTE Client provides:
- Test Manager for configuration of tests and test environment
- Automated test execution
- An API to create simulated test components (models)
- Managing of tests (unit tests, scripts, etc.)
- Real-time test result monitoring, recording, and playback
- Test result reporting and coverage
The OTE Test Server provides:
- A test execution manager
- A simulated test environment that manages/executes test models
- Managing I/O connections to the target test environment
- Managing the test environment resources
- Relay real-time test data to client
OSEE Functional Use Cases
OSEE is designed to work in a collaborative environment, supporting many developers working on the same data. One of the many considerations to help accomplish this is the use of a multi-tier architecture to maintain a common set of underlying data for all. Specifically, OSEE utilizes a three tier paradigm to separate the user interface from the component capabilities. There are three cases that are important to understand.
Single User Deployment
This is the default deployment if you download OSEE from the web. The embedded HSQL database will be automatically created when the server is run in a command shell. The client is configured to connect to the local server by default.
Multi-User, Single Application Server Deployment
The multi-user deployment can be achieved by installing OSEE clients on separate machines that have network access to the server machine. The client .ini file needs to be configured to connect to the Application Server machine. A more robust database can be configured for the Application Server via the database JOSN configuration file.
Multi-User, Multi-Application Server Deployment
Multi-Application Server deployment should be considered if:
- higher availability for the application server is required
- the Application Server becomes over burdened, impacting server response time
In these cases, load balancing across multiple application servers is encouraged. OSEE Application servers are designed to support load balancing, and do not require the client to connect to a specific application server, and do not maintain lists of the clients they service. The load balancer is allowed to distribute client requests according to its configured distribution scheme.
Images were created using Inkscape