SAFe: one option to scale Agile

The last year has been marked by the steady adoption of SAFe (Scaled Agile Framework) in companies that, although already development powerhouses, still had been struggling to transform themselves into Agile development shops. So SAFe has played an increasingly important role in those companies transitioning from more traditional SDLCs to Agile based ones.

So what is SAFe? It is the evolution of Dean Leffingwell's long time work in methodologies in general, starting with RUP, tempered with his practical experience in bringing Agile and Lean to many companies over the last ten years. According to its website, "The Scaled Agile Framework is an interactive knowledge base for implementing Agile practices at enterprise scale." As such is has a nicely crafted big picture with hundreds of icons you can drill down to and learn more about a specific topic, sometime going even deeper. It reminded me of the richness you get from the RUP documentation, albeit all of it related to current and relevant Agile topics.

image

The navigation is a bit bumpy because in order to protect their IP, the authors made it difficult to see more than one topic at a time (no “open link in new tab option”), and you can’t select text either, so if you are quoting them you will need to type it all. Other than that it can be seen as a rich “map” to all things Agile in an enterprise. The main references of SAFe are Leffingwell's books (both available at Safari):

How is SAFe different from Scrum? The short answer is that it encompasses Scrum. You could say that Scrum targets the team, and that SAFe targets the enterprise. Though based on Lean and Agile principles at the micro, team level, it also addresses architecture, integration, funding, governance and roles at the macro or enterprise level.

And here is what is relevant to a Business Analyst: it clearly provides a way to structure and manage requirements at the enterprise level. It starts by reasserting what has become pretty much the standard best practice in requirements management hierarchy levels, Portfolio, Program and Team:

· Portfolio: Business and Architecture epics, epics span releases

· Program: Features fit releases

· Team: Stories fit in iterations

The novelty of SAFe stems from providing practical solutions to some of the “elephant-in-the-room” problems that have prevented scaling up requirements management in many companies. Among these, three caught my attention:

· Architecture runways: this is something obvious for the technical team, but SAFe makes it explicit by adding Architecture epics as an item in the Portfolio backlog. An Architectural Runway “is the extant technical infrastructure (instantiated in code) necessary to support the implementation of upcoming features without excessive, delay-inducing, redesign.” By having Architecture topics explicit, businesses now become aware of the hidden part of the iceberg needed to implement business features: a plane can’t safely fly without a runaway.

· Agile Release Train: it is “a long-lived team of agile teams, typically consisting of 50-125 individuals, that serves as the program-level value delivery mechanism in SAFe. “ This is the equivalent of a Scrum team but at the program or release level. By having this concept clearly understood, business becomes aware about the need to support using a common team sprint cadence, and “teams are aligned to a common mission via a single Program Backlog”. You could say that this idea is similar to Scrum-of-Scrums concept, but it goes beyond in unifying the teams around the idea of releasing Potentially Shippable Increments (PSIs) as a unit, and not just synchronizing the efforts of separate Scrum teams at the backlog level but still not coordinating the delivery of working software, which is after all the measure of progress for Agile teams.

· Investment Themes: by adding this to a portfolio backlog, and mapping it directly downstream to its ramifications at program and team level, it ensures that what is being worked on has been budgeted, and that execution is tied to strategy. Business becomes aware of the need to prioritize and the repercussions downstream of short-changing initiatives that might affect its financial future.

Reception

SAFe is not without its critics. It has been bashed by both Lean and Agile/Scrum champions such as David Anderson, Ken Schwaber, and Mike Cohn. However it seems that those initial reactions "throw the baby out with the water", to use an old expression. There is definitely value in the framework, but only if considered within an understanding of how ideas are adopted and used within the enterprise.

For the Scrum practitioner, a minor annoyance is its confusing definition of Scrum: Leffingwell invented “SAFe ScrumXP” as a combination of Scrum project management techniques with XP engineering practices. This separation stems from a backwards pseudo-definition of Scrum from 15 years ago, when the perception was of Scrum as being solely a project management framework. At the time Mike Beedle coined the name “Xbreed” (later “Enterprise Agile Process”) to mean exactly Scrum + XP.

“Xbreed” didn’t catch because eventually everyone started to use the word “Scrum” to mean engineering practices as well (notice how Beedle already says that here even for XBreed), and later Scrum.org and ScrumAlliance made it official by adding the “Professional Scrum Developer” and “Certified Scrum Developer” courses focused exactly on teaching engineering practices alongside the Scrum project management framework.

Finally, among the strongest criticisms is that SAFe does not conform to a well-known process adoption best practice popularized by Alistair Cockburn: "stretch to fit". Like RUP, SAFe needs to be tailored in size, and many might be tempted to adopt everything when in doubt about what to do. SAFe can then be used at the Shu and maybe a bit at the Ha levels (as referred by Cockburn), but for experienced Agile practitioners, both in the technical and product management sides of the business, SAFe will appear overwhelming and verbose, and might get in the way as feeling over prescriptive. However, for companies coming out traditional SDLC middle ages, SAFe can feel like the map of a gold mine. It gives to the uninitiated in the Agile wisdom a sense of direction and security to finally start experimenting on how to get out of the Waterfall corner many businesses brought themselves to.

Some references pro and against:

Method Wars: Scrum vs SAFe, by Ian Mitchell

“SAFe has gained traction [with big companies] not in spite of poor agile credentials, but rather because of them.”

Has SAFe Cracked the Large Agile Adoption Nut, by InfoQ

“However, not all in the community think SAFe is a good idea. In fact, many have a strong negative reaction”

Controversy around SAFe, DAD and Enterprise Scrum, by Elizabeth Woodward

“SAFe has been empirically derived from addressing problems as teams scale--lessons learned over time--and it continues to evolve.”

unSAFe at any speed, by Ken Schwaber

“The boys from RUP (Rational Unified Process) are back.”

Kanban – the anti-SAFe for almost a decade, by David Anderson

“SAFe appears to collect together a number of techniques from software development processes from the 1990s and 2000s”

Comments are closed

Calendar

<<  March 2024  >>
MonTueWedThuFriSatSun
26272829123
45678910
11121314151617
18192021222324
25262728293031
1234567

View posts in large calendar

Month List