Although the global success of software is unquestionable (“software is eating the world”, everyone wants to use agile, etc.), last year the agile guru and “father” of microservices Martin Fowler was talking "why are we struggling" (see https://martinfowler.com/articles/agile-aus-2018.html ). He pointed at three problems:
- Get rid of the Agile Industrial Complex and the idea of imposing stuff on teams. Let teams work out the way they should work themselves.
- Raise the importance of technical excellence, and never forget that when writing software, the technology side is really vital, and
- Organize around products.
The point 1 is actually well-known for architecture – no organisation is using only one framework and a framework “as is”. It is always a mixture of several frameworks which have been “massaged” to handle the real and unique needs of an organisation. To overcome this problem, the serious of blogs “Better Architecting With” (BAW) https://improving-bpm-systems.blogspot.com/search/label/%23BAW introduces a different approach – architecture description (remember ISO/IEC/IEEE 42010) which is a set of views and models, these models are aligned, and some models are, also, parts of system-of-interest. Thus teams may use views and models as they want.
There is already a standard set of model-types (e.g. I use 100+ of them) and there are some methods to generate (sometimes semi-automatic) some models from existing models. Processes are not fixed, but there is a natural order of dependencies between model-types. So, the teams may have a lot of freedom by using standard methods and model-types.
The item 3 is also well-known for architecture – the entity (product, solution, system, etc.) life cycle is more important than projects at various stages of such life cycle (see https://improving-bpm-systems.blogspot.com/2018/01/better-architecting-with-explicit.html ). So, there is a clear separation of duties:
- Use architecture to define the units-of-work to be done and some dependencies between them.
- Use agile to schedule and monitor execution of such units-of-work.
It seems that the guru is “self-locked” in the software domain in accordance with the well-known observation “if all you have is a hammer than everything looks like a nail”. Software has its artefacts (classes, packages, interfaces, schemas, services, applications, platforms, etc.), however solutions are, also, built from a different set of artefacts (which use software artefacts, of course).
2 Solution artefacts
Architecturally, solutions consist of solution artefacts such as:
- UI forms,
- automation scripts,
- data structures,
- IoT devices,
Also, architecturally, each of these artefacts may have several facets:
- specific management tool, e.g. Business Process Management (BPM)-suite tool;
- templates, e.g. process templates;
- instances, e.g. process instances;
- APIs (or interfaces) to all functionality, and
- patterns used by templates, e.g. workflow patterns.
Logically, all the units-of-functionality to be shared should form an enterprise-wide platform (this is a type of software artefacts) and some units-of-functionality may become microservices (this is a type of software artefacts).
3 Platforms and microservices orientation
The logic behind platforms is rather straightforward – instead of developing traditional applications which contain a lot of duplicating functionally, all common functionality is collected into a platform and solutions which are built on top of that platform possess only unique functionality. See figure below.
Planform is designed on the following principles:
- The platform simplifies the use of its platform components and ensures interaction between them. Thus, the platform frees up resources to focus on solving unique problems.
- New solutions are implemented outside the platform using rapid development techniques (agile methodology) and platform existing functionality.
- The planning of new solutions is coordinated within the scope of the platform to minimize duplication of efforts in solving the same problems.
- Successful new (thus innovative) solutions are gradually included in the platform for their wide use.
- Constant improvement of the platform is achieved through transparency, feedback, results of use and systematic evaluation of existing platform components.
- All interactions inside and outside the platform use the standardized interfaces (API methodology).
- The platform provides for several interchangeable options for some platform components.
- The platform offers not only a rich functionality, but also a methodology for developing various solutions based on it.
- The platform is built gradually by increasing its functionality horizontally and vertically.
The platform-enabled agile solutions orientation implies common engineering practices for solutions created on top of the platform. Any solution which is built on such a platform must contain only functionality which is not implemented in the platform (see Figure below).
4 Smart cities reference architecture case
Let us illustrate platforms, solutions and solution artefacts with the Smart Cities Reference Architecture Methodology which is under development in the IEC System Committee “Smart Cities”. Some of the goals of this methodology are to guarantee that
- all the common functionality is developed once and used by many Smart Cities and
- individual Smart City can easily developed their own unique solutions (including replacing some platform functionality).
The analysis of Smart Cities functionality demonstrated that minimum 3 platforms are necessary: universal, urban-generic and urban-specific (see Figure below).
Another option is to split the urban-specific platform into 14 platforms.
To facilitate the isolation of solution artefacts, a reference application architecture for solutions is proposed (see figure below).
Then all the various facets (tools, APIs, patterns, templates and instances) are mapped to the platforms and solutions (see figure below).
5 Typology of solutions
The reference application architecture for solutions can be further elaborated by adding a typology (classification) of solutions. This means that there are some types of solution and each type is formed around a predominant solution artefact, namely:
- event centric,
- data-entry centric,
- document/content centric,
- data or/and information flow centric,
- data or/and information visualisation,
- IoT-device centric,
- short-running operations (or activity-based),
- long-running operations (or process-based),
However, a real solution, usually, combines some of those types.
Thus, solutions are delivered in accordance with the following generis procedure:
- Minimal architecting to understand the type (or a mixture of types) of a solution.
- Collecting use cases to define capabilities of the solution.
- Quick prototyping to outline functions, APIs, and solutions artefacts.
- Conducting gap analysis to determine what is missing in the platform(s).
- Developing missing functionality as microservices to close the gap.
- Assembling newly created and existing microservices to deliver the solution.
And, don’t forget the versioning for everything.
This blogpost belongs to the series https://improving-bpm-systems.blogspot.com/search/label/%23BAW