This is the lesson planted at 1:33 and paid off here: boring technology, implemented well, beats new technology (15:21).
The complexity budget
- Every team has a finite budget of complexity it can afford to manage (
15:50). - Every new piece of infrastructure eats into that budget, because you must (
15:50):- learn its quirks
- build operational expertise around it
- train your team
- hire for it
- These costs are very real even though they never show up on a balance sheet (
15:50).
Why boring tech is nearly free
Postgres, Linux, Memcached, Nginx cost almost nothing to adopt because (16:13):
- The team already knows them.
- The internet is already full of answers to your problems.
- Failure modes are documented in books.
- The operational tooling has existed for 15+ years.
When you should spend the complexity budget
Spend it only when you have a genuinely new problem the boring stuff cannot solve (16:13):
- Vector search at scale
- Real-time time-series analytics
- Graph traversal at planet scale
Those are categories where specialized databases are genuinely required (16:45).
"A lot of users" is not a new problem
- Having many users is not novel — every system at scale has a lot of users, and Postgres can handle a lot of users (
16:45). - Proof points cited: Instagram, Notion, Reddit, GitHub (for many of its workloads), and Stripe at some capacity (
16:45). - The core diagnosis of the video (
16:45):
The mistake most teams make is not picking the wrong database — it's switching databases too early, when the real fix was just to understand their existing database better.