Choosing the right SDLC (Software Development Lifecycle) methodology for your project is as important to the success of the project as the implementation of any project management best practices. Choose the wrong software methodology and you will add time to the development cycle. Adding extra time to the development cycle will increase your budget and very likely prevent you from delivering the project on time.
Choosing the wrong methodology can also hamper your effective management of the project and may also interfere with the delivery of some of the project's goals and objectives. Software development methodologies are another tool in the development shop's tool inventory, much like your project management best practices are tools in your project manager's tool kit. You wouldn't choose a chainsaw to finish the edges on your kitchen cabinet doors because you know you wouldn't get the results you want. Choose your software methodology carefully to avoid spoiling your project results.
I realize that not every project manager can choose the software methodology they will use on every project. Your organization may have invested heavily in the software methodology and supporting tools used to develop their software. There's not much you can do in this case. Your organization won't look favorably on a request to cast aside a methodology and tools they've spent thousands of dollars on because you recommend a different methodology for your project. We'll give you some tips on how to tailor some of the methodologies to better fit with your project requirements later in this article. In the meantime, before your organization invests in software development methodologies you, or your PMO, ought to be consulted so that at least a majority of projects are benefited from a good fit.
This article won't cover every SDLC out there but we will attempt to cover the most popular ones.
Scrum
Scrum is a name rather than an acronym (which is why I haven't capitalized the letters), although some users have created acronyms, and is commonly used together with agile software development. Scrum is typically chosen because of its iterative nature and its ability to deliver working software quickly. It is chosen to develop new products for those reasons. There is typically no role for a project manager in this methodology, the 3 key roles are: the scrum master (replacing the project manager), the product owner, and the team who design and build the system. There is only one role that you would be asked to play if your organization is committed to using this methodology, scrum master. If you should determine that this would actually be the best methodology for your project, you'll have to re-examine your role as project manager. You can either identify a suitable scrum master and return to the bench, or fill the role of scrum master.
Scrum suits software development projects where its important for the project to deliver working software quickly. Scrum is an iterative methodology and uses cycles called sprints, to build a working system. Requirements are captured in a "backlog" and a set of requirements is chosen with the help of the product manager. Requirements are chosen based on 2 criteria: the requirement takes priority over others left in the backlog and the set of requirements chosen will build a functioning system.
During the sprint, which can last from 2 to 4 weeks maximum, no changes can be made to the requirements in the sprint. This is one of the reasons that a project manager isn't necessary for this methodology. There is no need for requirements management because no changes are allowed to the requirements under development. All changes must occur in the requirements set in the backlog.
Scrum will be suitable for software development projects where the product is a new software product. By new I mean that it is new to the organization undertaking the project, not in general. The methodology was developed to address a need for a method to build software when its necessary to learn on the fly, not all requirements are known to the organization and the focus is on delivering a working prototype quickly to demonstrate capabilities. You need to be careful when choosing requirements to deliver in each sprint to ensure that the set developed builds a software system that is capable of demonstrating the feature set supporting the requirements included.
You also need to ensure that these requirements are well known and understood as no changes are allowed once the sprint starts. This means that any changes to the requirements must come through a new set of requirements in the backlog making changes to these requirements very expensive.
This methodology divides stakeholders into 2 groups: pigs and chickens. The inventors of this methodology chose this analogy based on the story of the pig and the chicken - it goes something like this. A pig and a chicken were walking down the road one morning and happened to notice some poor children who looked like they hadn't eaten for days. The compassionate chicken said to the pig: "Why don't we make those children a breakfast of ham and eggs?" The pig said: "I'm not happy with your suggestion. You're just involved in making the breakfast, I'm totally committed!" The point to this is the product owner, scrum master, and team are all in the "pig" group. All others are in the "chicken" group. You will be in the "chicken" group if you choose the Scrum methodology as a project manager.
Waterfall
Waterfall methodology calls for each phase of the development cycle to be repeated once only. Requirements will be gathered and translated into functional specifications once, functional specifications will be translated to design once, designs will be built into software components once and the components will be tested once. The advantage of this methodology is its focus. You can concentrate the effort of all your analysts on producing functional specifications during one period rather than have the effort dispersed throughout the entire project. Focusing your resources in this way also reduces the window during which resources will be required. Programmers will not be engaged until all the functional specifications have been written and approved.
The disadvantage of this approach is its inability to teach the project team anything during the project. A key difference between the waterfall approach and an iterative methodology, such as Scrum or RUP, is the opportunity to learn lessons from the current iteration which will improve the team's effectiveness with the next iteration. The waterfall methodology is an ideal methodology to use when the project team has built software systems very similar to the one your project is to deliver and has nothing to learn from development that would improve their performance. A good example of a project which would benefit from the waterfall methodology is a project to add functionality to a system the project team built in the not too distant past. Another example of an environment that is well suited to the waterfall methodology is a program to maintain a software system where a project is scheduled for specific periods to enhance the system. For example, an order and configuration software system which is enhanced every 4 months.
The waterfall methodology does not lend itself particularly well to projects where the requirements are not clearly understood at the outset. Iterative approaches allow the product owners or user community to examine the result of building a sub-set of requirements. Exercising the sub-set of requirements in the iteration's build may cause the product owners or user community to re-examine those requirements or requirements to be built. You won't have that opportunity with the waterfall method so you need to be certain of your requirements before you begin the build phase. Interpreting requirements into functionality is not the only aspect of development that can benefit from an iterative approach. Designing the system and building it can also benefit from doing these activities iteratively. You should use the waterfall method when your team is familiar with the system being developed and the tools used to develop it. You should avoid using it when developing a system for the first time or using a completely new set of tools to develop the system.
RUP
The Rational Unified Process, or RUP, combines an iterative approach with use cases to govern system development. RUP is a methodology supported by IBM and IBM provides tools (e.g. Rational Rose) that support the methodology. RUP divides the project into 4 phases:
1. Inception phase - produces requirements, business case, and high level use cases
2.Elaboration phase - produces refined use cases, architecture, a refined risk list, a refined business case, and a project plan
3. Construction phase - produces the system
4. Transition phase - transitions the system from development to production
RUP also defines 9 disciplines: 6 engineering disciplines, and 3 supporting disciplines: Configuration and Change Management, Project Management, and environment so is intended to work hand in hand with project management best practices.
Iteration is not limited to a specific project phase - it may even be used to govern the inception phase, but is most applicable to the construction phase. The project manager is responsible for an overall project plan which defines the deliverables for each phase, and a detailed iteration plan which manages the deliverables and tasks belonging to each phase. The purpose of the iterations is to better identify risks and mitigate them.
RUP is essentially a cross between Scrum and waterfall in that it only applies an iterative approach to project phases where the most benefit can be derived from it. RUP also emphasizes the architecture of the system being built. The strengths of RUP are its adaptability to different types of projects. You could simulate some of the aspects of a Scrum method by making all 4 phases iterative, or you could simulate the waterfall method by choosing to avoid iterations altogether. RUP will be especially useful to you when you have some familiarity with the technology but need the help of Use Cases to help clarify your requirements. Use Cases can be combined with storyboarding when you are developing a software system with a user interface to simulate the interaction between the user and the system. Avoid using RUP where your team is very familiar with the technology and the system being developed and your product owners and users don't need use cases to help clarify their requirements.
RUP is one of those methodologies that your organization is very likely to have invested heavily in. If that's your situation, you probably don't have the authority to select another methodology but you can tailor RUP to suit your project. Use iterations to eliminate risks and unknowns that stem from your team's unfamiliarity with the technology or the system, or eliminate iterations where you would otherwise use the waterfall method.
JAD
Joint Application Development, or JAD, is another methodology developed by IBM. It's main focus is on the capture and interpretation of requirements but can be used to manage that phase in other methodologies such as waterfall. JAD gathers participants in a room to articulate and clarify requirements for the system. The project manager is required for the workshop to provide background information on the project's goals, objectives, and system requirements. The workshop also requires a facilitator, a scribe to capture requirements, participants who contribute requirements, and members of the development team whose purpose is to observe.
JAD can be used to quickly clarify and refine requirements because all the players are gathered in one room. Your developers can avert misunderstandings or ambiguities in requirements by questioning the participants. This method can be used with just about any software methodology. Avoid using it where the organization's needs are not clearly understood or on large, complex projects.
RAD
RAD is an acronym for Rapid Application Development uses an iterative approach and prototyping to speed application development. Prototyping begins by building the data models and business process models that will define the software application. The prototypes are used to verify and refine the business and data models in an iterative cycle until a data model and software design are refined enough to begin construction.
The purpose of RAD is to enable development teams to create and deploy software systems in a relatively short period of time. It does this in part by replacing the traditional methods of requirements gathering, analysis, and design with prototyping and modeling, the prototyping and modeling allow the team to prove the application components faster than traditional methods such as waterfall. The advantage of this method is it facilitates rapid development by eliminating design overhead. It's disadvantage is that in eliminating design overhead it also eliminates much of the safety net which prevents requirements from being improperly interpreted or missed altogether.
RAD is suitable for projects where the requirements are fairly well known in advance and the data is either an industry or business standard, or already in existence in the organization. It is also suitable for a small development team, or a project where the system can be broken down into individual applications that require small teams. RAD is not suitable for large, complex projects or projects where the requirements are not well understood.
LSD
Lean Software Development, or LSD, applies the principles of waste reduction from the manufacturing world to the business of developing software. The goal of LSD is to produce software in 1/3 the time, on 1/3 the budget, and with 1/3 the defects of comparable methods. Lean does this by applying 7 principles to the endeavor of software development:
1. Eliminate waste
2. Amplify Learning (both technical and business)
3. Decide on requirements as late as possible
4. Deliver as fast as possible
5. Empower the team
6. Build integrity
7. See the whole
Although Lean Manufacturing has been around for some time, its application to the process of developing software is relatively new so I wouldn't call it a mature process.
LSD would be a suitable method to use where you have a subject matter expert in the method who has some practical experience in applying lean methods to a software development project. "Amplified" learning implies that your development team has a depth of knowledge in the software tools provided, and also a breadth of knowledge that includes an understanding of the business needs of the client. LSD would be suitable for a project where the development team has these attributes.
LSD depends on a quick turnaround and the late finalization of requirements to eliminate the majority of change requests, so will not be suitable for a project where a delayed finalization of requirements will have a poor chance of eliminating change requests, or the size and complexity of the system being developed would prevent a quick turnaround.
Extreme Programming (XP)
Extreme programming places emphasis on an ability to accommodate changes to requirements throughout the development cycle and testing so that the code produced is of a high degree of quality and has a low failure rate in the field. XP requires the developers to write concise, clear, and simple code to solve problems. This code is then thoroughly tested by unit tests to ensure that the code works exactly as the programmer intends and acceptance tests to ensure that the code meets the customer's needs. These tests are accumulated so that all new code passes through them and the chances for a failure in the field are reduced.
XP requires the development team to listen carefully to the needs and requirements of the customer. Ambiguities will be clarified by asking questions and providing feedback to the customer which clarifies the requirements. This ability implies a certain degree of familiarity with the customer's business; the team will be less likely to understand the customer's needs if they don't understand their business.
The intent of XP is to enhance coding, testing, and listening to the point where there is less dependency on design. At some point it is expected that the system will become sufficiently complex so that it needs a design. The intent of the design is not to ensure that the coding will be tight, but that the various components will fit together and function smoothly.
XP would be a suitable software development method where the development team is knowledgeable about the customers business and have the tools to conduct the level of testing required for this method. Tools would include automated unit testing and reporting tools, issue capture and tracking tools, and multiple test platforms. Developers who are also business analysts and can translate a requirement directly to code are a necessity because design is more architectural than detail. This skill is also required as developers implement changes directly into the software.
XP won't be suitable where the development team does not possess business analysis experience and where testing is done by a quality assurance team rather than by the development team. The method can work for large complex projects as well as simple smaller ones.
There is no law that states you must choose one or the other of these methodologies for your software project. The list I've given you here is not a totally comprehensive list and some methodologies don't appear on it (e.g. Agile) so if you feel that there is some other methodology that will better suit your project, run with it. You should also look at combining some of the features of each of these methods to custom make a methodology for your project. For example, the desire to eliminate waste from the process of developing software is applicable to any method you choose and there is likely waste that could be eliminated in any development shop.
Be careful to choose a methodology that is a good fit for your team, stakeholders, and customer as well as your project. Bringing in a new development methodology that your team will struggle to learn at the same time they are trying to meet tight deadlines is not a good idea. On the other hand, if you have the latitude you may want to begin learning a new method with your project.
Posts
