Recent texas flash flooding killed 50+ people

yesterday i read many of my old posts when migrating them to github

I found I mentioned busy and potential burnout lots of times

i also wrote i should think about simplify, reduce complexity etc.

but seems i always in "drown" mode due to large piles of work, endless meetings, outages, urgent deadlines etc

considering the texas flash flood, i need to think about this again

okrs, project deadlines, review requests, emails, chats, questions, outages, oncall alerts

they are all like water, big or small

and they can quickly pile up

we are a processor, or a big tunnel, or a river bed

we have limited capacity, throughput, latency

like flooding, if the throughput is not large enough, water will overflow

and we are drown

we need to think about ideas to mitigate / prevent overflow

  1. start with okr. we should be cautious about optimistic time estimation. and should consider the throughput of project owner.
  2. automation. for the same kind of tasks / issues, automate as much as possible.
  3. oncall rotation. and reduce false positives
  4. for consulting questions, establish standard q&a channel, like group email / bugs, assign appropriately, avoid most questions going to a single person. also create docs and potentially ai bot/agent.
  5. for projects, simplify the design, go lightweight.
<
Previous Post
Agent programming lesson 5 - testing
>
Next Post
Resource integration with llm