Fork me on GitHub
Daniel Hines00:10:17

@lady3janepl, that is a good idea, one k hadn't considered.


what @lady3janepl said - SLAs and SLOs are super important use things like micro APIs and functions as a service to control your billing, keep track of invocations etc


use IaaS if you're people-limited, use infrastructure you run yourself if you're liquid cash limited


get your deployments in a zero-downtime state ASAP, even if it's costly to do


make sure you can always release the tip of master


if you can't, halt work elsewhere and fix that


a dollar spend on good CI/CD will come back to you twice over


aaaand finally, my friend who's an amazing product manager and has run businesses before says that with anything 'execution is everything'


so yeah the idea, the tech, everything doesn't matter if it's not packaged right - which includes the sell, the audience, the marketing, and the story you tell about the product & how it makes users feel


And a bit of explanation for the above... Speaking as a consumer of APIs, SLA/SLO is the first thing I think about (even before docs). Nothing pisses me off more than company being shifty about their downtimes, because this makes me look incompetent in the eyes of my customers. “Sorry guys, I couldn’t tell you we’d be having downtime because my provider doesn’t warn us.” “THEN SORT YOUR PROVIDER YOU IDIOT”

👍 16
💯 4

Service downtimes work exactly like cars braking down on the motorway. There’s a delay in reaction and the car speed loss increases non linearly with the amount of cars participating in the traffic jam.


So when you’re writing software for end users you’re relatively safe, but when you’re writing something B2B (like an API gateway, grrr) you can have significant impact on a chain of companies.