Day 15 results: a dict in front of a slow database

Measured on: (machine, OS, Python version)

Output

Paste part 1, part 2, part 3 and the scoreboard here.

One sentence per prediction

P1 (cache hit vs source read, the speedup floor):
P2 (cache-aside hit rate on the Zipf workload):
P3 (source load reduction, no cache vs cache-aside):
P4 (load multiplier when hit rate drops 90% to 80%):

The mechanism, in my own words

Walk the cache-aside read path for a hit and for a miss. On a miss, what two things happen before the value comes back:

Why the cache changed everything

The hit rate I measured, the database calls with and without the cache, and how much faster the average read got:

The non-linear bit that surprised me

Going from a 90% hit rate to 80% did what to database load, and why is that not a 10% change:

The number that matters

A cache hit was __ times faster than a source read. At a __% hit rate the cache cut database calls by ___x. Dropping to 80% multiplied database load by ___x. One line on why hit rate, not raw speed, is the thing to watch: