Andrew McVeigh at #qconsf: Riot realized they needed to decompose monolith after a multi day outage. I think that is a common trigger.
1
5
17
I’ve had similar experiences in several organizations - what both sales and ops people would call a “compelling event”. But subtly, it’s not that the org all of a sudden realizes; it’s that the org can now *justify* the effort.
1
10
That's an important point.
1
2
Replying to @mtnygard @randyshoup
Interestingly, the most common trigger I observe is not runtime problems, but development time issues. One major reason for different modularization strategies

Nov 6, 2018 · 6:57 AM UTC

2
1
4
Replying to @stilkov @mtnygard @randyshoup
1/ imho we find multiple triggers: increasing time-to-market, new features take too long to build Reasons: * low inner quality = technical debt, * bad technology choices, * to many dependencies (aka bad modularization/cohesion) * fundamentally flawed development process
2
2
Replying to @stilkov @mtnygard @randyshoup
2/ increasing nr of runtime-issues, i.e.: * with high financial impact for business * impacting large number of users * paralyzing the support organization