compute/stats-for-sysadmins.md
... ...
@@ -2,13 +2,17 @@
2 2
3 3
### 1/n chance tried n times
4 4
From https://theartofmachinery.com/2020/01/27/systems_programming_probability.html
5
+
5 6
For n>3, there's a bit less than a 2/3 chance of at least one hit.
6 7
7 8
### n balls in n buckets
8 9
From https://theartofmachinery.com/2020/01/27/systems_programming_probability.html
10
+
9 11
If you throw n balls into n buckets at random, a bit over a third of the buckets will be empty, a bit over a third will have one ball, and the remaining quarter or so will have multiple balls.
10 12
11 13
### Scale up, or scale out?
12 14
From http://people.brunel.ac.uk/~mastjjb/jeb/or/queue.html
15
+
13 16
If you only care about throughput, scale up.
17
+
14 18
If you care about wait times, scale out.