,
5 tweets,
1 min read
Read on Twitter
i've definitely described honeycomb as "strace for microservices" more than once.
but ironically, i think that the era of most-engineers-needing-access-to-system-level-resources is in its final throes.
if you're writing app code, you shouldn't have to care about anything below the level of functions and APIs (and distsys hygiene around timeouts/retries).
if you're writing app code, you shouldn't have to care about anything below the level of functions and APIs (and distsys hygiene around timeouts/retries).
anything lower level than that is a specialist skill, increasingly relegated to the specialist engineering teams of cloud platform services and db services.
(have i ranted about this recently on twitter? i don't think so)
(have i ranted about this recently on twitter? i don't think so)
tldr the ops profession has been the human buffer connecting low level systems and software engineering for decades. this era is coming to an end.
systems are now too large and complex to be cached in anyone's head, and quality user experience demands end-to-end sw ownership.
systems are now too large and complex to be cached in anyone's head, and quality user experience demands end-to-end sw ownership.
software engineers looking to the future should consider how folks work on 'serverless' (sic) code and systems. i.e. assume no underlying knowledge of the system internals, experience reality purely thru the lens of instrumentation.
as goes serverless, so goes the profession.
as goes serverless, so goes the profession.
