Source Resources
- AaaS - Architecture-as-a-Service
- many competing agendas
- technology change is a constant
- nobody wants to be on legacy platform
- java foundational tech for aws, google, twitter, netflix, etc.
- how to evaluate technology
- chasing the new things
- computer science - lifetime learning
- education doesn’t end. lifetime of learning.
- developers have strong opinions
- new and old tech https://youtu.be/d5bNZX8tpiI?t=681
- new - unfinished, unbuggy, unproven
- old - refined, stable, tested
- predictable hype cycle
- how do we know where not to use tech. takes trial and error.
- asked to build something never built before, using tools we got a few weeks ago, shocked we don’t know how long it takes
- developer get bored quickly
- learning keeps it fresh
- commit at some point. can’t constantly experiment
- bleeding edge … means you will bleed
- pioneers … ones with arrows in back
- hope is not a strategy
- need to be deliberate when it comes to technology choices
- challenge is how to keep up (“technology merry go round”)
- make it part of routine to learn
- “morning coffee” - peruse the tech news
- attention is most precious resource you have
- attention is a resource. it doesn’t scale.
- don’t waste your attention. have to be selective.
- if its a big enough deal, you will hear about it.
- how do we know where to invest our time?
- be aware of bias
- skating to where the puck was
- thoughtworks technology radar
- 20% projects are good space, but fallen out of favor in some circles
- innovation fridays
- architectural briefing - one person does research, presents back to team. short ~45min
- briefing pans out -> workshop/invest more in it
- greenfield always works. need to use in target env to understand constraints
- start with low risk capability / not company core
- security - want to stay safe, need to go fast. time feeds black hats
- policy needed. measure and enforce, hard part
- pros and cons - always trade-offs
- “it depends” means -> tell me more. not a stopper.
- ack the negatives. the most insight is learned from this question
- “You haven’t mastered a tool until you understand when it should not be used” - Kelsey
- what is the appropriate scale for evaluation. e.g. 1-5
- harvey balls. images to remove the “hard number” focus effect
- nobody tells you which criteria to use, up to you, need to make with incomplete information
- architects spread thin, need to scale with principles, north stars, etc.
- arch agile backlash.
- chose tech that can evolve - evolutionary arch
- create arch with expectation of change. à la werner “everything fails, all the time”
- “evolutionary arch supports guided, incremental change across multiple dimensions”
[arch]
fitness functions0: make arch change, 1: eval result, 2: closer to goal? yes -> keep, no -> trash, 3: -> goto 0
(basic regression iterations as in stats/ML)- set of tests we exec to validate arch. (arch unit tests)
- reminds everyone on the team what’s important for your arch
- informs discussions around tradeoffs
- need to be visible to team
- need to be reviewed on regular basis
- just because you can measure it doesn’t mean it matters
- avoid resume driven design
Twitter • Reddit