Blog
Notes on methodology, infrastructure, and what the data shows.
-
03 Nov 2026
October 2026 in data: the longest outage was on the calendar
46 services, 179 lost minutes. Most belong to one Linode outage announced a month in advance, which still ran past both ends of its window.
data reliability incidentsRead -
08 Sept 2026
The endpoint you monitor decides the outage you can detect
An audit of our leaderboard found 17 of 50 services that could not have recorded an outage. Not a bug: a reasonable rule meeting a badly chosen URL.
methodology monitoring reliabilityRead -
11 May 2026
What 99.99% uptime actually buys you
Reading cloud SLAs: the math behind "the nines", four ways vendors quietly reshape the denominator, and what to look at instead of one headline number.
sla reliability methodologyRead -
08 May 2026
Why independent uptime measurement matters
Provider status pages are structurally unreliable. Here is what independent measurement looks like and why it matters for engineers picking infrastructure.
announcement methodologyRead