Share
Embed
Michael DeHaan - Keynote - Speaking for the Dead: Is "Waterfall" and "Monolithic" Actually Good?
devopsdaysraleigh
October 06, 2016
Transcript
25 slides
-
Speaking for the Dead: Is "Waterfall" and "Monolithic" Actually Good?
MICHAEL DEHAAN @LASERLLAMA - 9.6.2016 -
PRELUDE u In Early 2012, I created a project called
Ansible. u Named after Ender’s Game u Orson Scott Card wrote “Speaker for the Dead” as a sequel -
PRELUDE u An honest, instructional, and empathetic eulogy. u I
began writing software full time at IBM in 2001 u I’m also not dead. u We’ve declared Waterfall and Monolithic Architecture dead. u Let’s talk about that. -
TODAY u “Waterfall” vs “Agile” u “Monolithic” vs “Microservices” u
Labels, spin, and assumptions u An appreciation of the very recent past -
First: A Theory u We get bored at work. u
We tend to like a challenge. u We hastily invent new things, that are not always better. u New people come into the field, and haven’t had the experience of older ways. u We get caught up in spin and “grandfalloons” (see Cat’s Cradle). u We assume there is a true way. There is no true way. -
Labels Apply Spin Good Connotation Bad Connotation Affordable Care Act
“Obamacare” USA PATRIOT ACT Agile Waterfall Microservices Monolithic Big Design Up Front Enterprisey -
TRUE NAMES u “Ansible” was originally coined by Ursula LeGuinn.
u In Earthsea, she writes of “True Names” u We must know things as they really are u “Waterfall” and “Monolithic” – do we know them, or just their given names? -
“Waterfall” and “Iterative Development” u In the early 2000s, waterfall
was the norm. u Requirements – Design – Implementation - Test u Engineering was engineering – greater responsibility, less hand holding. u Plenty of time to design solutions and research. -
Waterfall and Iterative Development (Cont) u Marketing Requirements Documents u
Engineering Requirements Documents u Questions, comments, documented mutual agreement u Status meetings usually once a week u Accurate view of strengths of various team members u Accurate view of dependencies between tasks, task order -
Agile (especially Scrum) u “Big Design Up Front” is labelled
“bad”. Is it? u Engineers are viewed as interchangeable. u Estimates are encouraged to not be time based u Scope is more unpredictable. u Overhead and status-focus is increased u Practices/Games make work feel like Kindergarten - “fist of five”, “planning poker” u Crunch time becomes constant -
Even In Lightweight Agile Systems u Task inter-relationships ignored u
Design time is shortened u Estimates are forced down u Requirements are more vague u No mutual sign-off. u Testing can suffer. -
A Middle Ground u Do share design u Do have
an accessible product owner, but a good one u Know hard requirements vs soft requirements. u Lightweight agile and Iterative Development can be closer than we think u Test driven development and test automation = super important. -
Example Worldwide Microservices Shop u ~150 services u Many thousands
of Jenkins builds per day u 8 Amazon regions (dev, stage, many prod regions) u ~1000s of EC2 instances u Could be modelled in ~50 EC2 instances in monolithic models u Jenkins configurations drift u Security enforcement drifts u Code sharing suboptimal u “DevOps” writing lots of custom code to manage it. -
Lessons Learned From Monoliths u Increased ability to develop on
one laptop u Interfaces can benefit from static typing vs design-by-contract u Very small numbers of different server/image types in production u Local function calls are much faster than RPC -
Lessons Learned From Monoliths (Cont) u Not many instances sitting
around with 5% utilization u No need for “containers” to get around above utilization problem u Greatly simplified debugging u Easy code sharing u Very simple version upgrades -
Microservices u Was created as a technical solution to a
people problem u Design by contract is effective u but increases test, ops, and debugging demands u Versioning increases need for communication and development formality u Avoids controversial design debates by … avoiding design u Potentially reduces code sharing, increases potential for NIH u Creates an infinite sea of new management challenges. -
Lessons Learned From Microservices u Design by contract enables: u
Language/tech choice u Documenting of interfaces u An API focus u Better decoupling from UI -
Conclusions u Labels cloud thought. u Monolithic architectures ARE more
efficient. u Agile can be less efficient. u Microservices attempts to solve a people problem by introducing new technical and people problems. u There’s something to learn from everything. u Try to separate the labels from the true meanings of things.