Starlink network health

AS14593 AS27277Normal

SquidGrid measures 17,181 internet paths that cross Starlink’s network, from vantage points in 42 countries, and compares each path with its own normal latency. Snapshot 2026-10-04 17:34 UTC.

Right now

83 of these paths have a reading from the last six hours. The typical path is running +0.0 ms against its own normal.

Change vs each path’s normal	Paths	Share
Under +25 ms (normal)	83	100.0%
+25 to +50 ms (elevated)	0	0.0%
+51 to +150 ms (high)	0	0.0%
+151 ms and up (severe)	0	0.0%
Unreachable (no route)	0

Routing: 0 of Starlink’s measured paths sit inside a prefix that was withdrawn from BGP or changed route in the last two hours.

The last 30 days

Each bar is one UTC day: the share of Starlink paths we sampled that spent at least one hour 25 ms or more over their normal. A typical day is 3.1% (dashed line); a bar turns orange at twice that.

Day (UTC)	Paths sampled	Over normal	Share	Average change
2026-10-03	2,080	45	2.2%	+6.3 ms
2026-10-02	2,180	72	3.3%	+6.8 ms
2026-10-01	2,661	59	2.2%	+5.9 ms
2026-09-30	2,005	51	2.5%	+6.5 ms
2026-09-29	1,926	115	6.0%	+7.2 ms
2026-09-28	2,036	131	6.4%	+7.4 ms
2026-09-27	2,033	112	5.5%	+7.3 ms

Days that stood out

2026-09-28 — 6.4% of measured paths spent an hour 25 ms or more over their normal, against a typical 3.1%.
2026-09-15 — 6.4% of measured paths spent an hour 25 ms or more over their normal, against a typical 3.1%.
2026-09-24 — 6.2% of measured paths spent an hour 25 ms or more over their normal, against a typical 3.1%.

Official status

No comprehensive official status feed. No public, network-wide status feed from Starlink is on record. Everything on this page is SquidGrid’s own measurement.

Is it Starlink, or the route to it?

A network-wide figure tells you whether Starlink as a whole is having a bad hour. It cannot tell you whether your traffic is affected, or where on the route it changes.

🔒 Customers

Which handoffs into and out of Starlink are off baseline, and in which cities

The carrier on each side of the stretch where conditions change

Alerts when your own sites’ paths cross a degraded stretch

Included with any paid plan.

See plans →

How to read this: a path’s normal is its own trended median latency, not a speed-of-light estimate. The bands follow SquidGrid’s state scale (+25 / +50 / +150 ms). The header reads Normal while under 2% of currently read paths are 25 ms or more over normal, Elevated from 2%, and Degraded from 10%. ASNs counted: AS14593, AS27277. Updated hourly; times in UTC. SquidGrid is not affiliated with Starlink.