use one big server

see also: Latency Budget · Platform Risk

scene cut

An essay argued that many systems should start with one big server rather than premature distributed complexity (source). The piece reframed scale as a sequencing decision.

signal braid

• Simplicity can outperform distribution early.
• Ops overhead grows faster than performance needs.
• Cost models favor vertical scaling in many cases.
• The ops angle matches Self-Hosting Email Ends.

single-line take

Most systems should earn their complexity.

link hop

This links to Self-Hosting Email Ends and State of HTTP in 2022 and Building a VM Inside ChatGPT.

open loop

What metrics should trigger the jump to distributed systems?