The Squid’s InkSquidGrid’s weekly read of the internet — fresh ink every Monday

Week 44, 2026 · Oct 19 – Oct 25

Covers Monday Oct 19 00:00 to Monday Oct 26 00:00 UTC.

How to read this: we measure millions of internet routes around the world. Every speed number below is how many milliseconds (thousandths of a second) slower a route ran than its own normal speed, so higher is worse. Each section says whether things got better or worse than last week.

2,722,815internet routes we measure
647,442,229measurements taken this week up 7% from last week
2,124outages and problems companies reported up 48% from last week
707of them rated major or critical

Bharti Airtel’s rough weekthe week’s story

Worse than last week. Bharti Airtel is one of the large networks that carry internet traffic between countries. Routes through it ran 32.5 milliseconds slower than normal on average this week (last week: 29.8). 1,663 routes through it were badly slowed (more than 25 milliseconds over normal) at some point. The chart shows how many, hour by hour:

Signed-in customers see the affected routes, the carrier on each side of every changed handoff, and the regional split on the live board.

The networks that slowed down the most

Large carrier networks, ranked by how much slower than normal their routes ran this week. Big networks get no allowance for their size.

Networkthis week, hour by hourslower than normalroutes badly slowedlast weekcompared with last week
SG.GS (Singapore network operator)45.9 ms93946.3 msAbout the same as last week
FLAG Telecom35.7 ms26435.8 msAbout the same as last week
Bharti Airtel32.5 ms1,66329.8 msWorse than last week
TransTelecom27.4 ms54027.0 msAbout the same as last week
Vocus Communications27.1 ms41427.5 msAbout the same as last week
Telxius25.0 ms1,814—New to this list this week

“Routes badly slowed” counts routes that ran more than 25 milliseconds over normal at some point this week.

2,124 outages and problems reported by companies

More than last week (up 48%). These are problems online services posted on their own status pages. Last week: 1,439.

SaaS1,363
Payments & fintech254
AI156
Education & research77
Cloud & hosting66
Crypto45
Consumer43
Gaming39

Online business software (SaaS) had the most problems again, as it usually does.

What tended to break, in providers’ own words: connectivity ×203 · slowness ×152 · auth & login ×98 · elevated errors ×91

Longest problems to fix: UNLV — Workday I-9 Form Informational Error (502 hours, about 21 days); NVIDIA GeForce NOW — Wuthering Waves issue (497 hours, about 21 days); Kaseya — DattoRMM - All Platforms - MacOS Agent Connectivity Issues (456 hours, about 19 days).

Ground eventsthe physical world, and whether the internet flinched

2026-10-21 A critical PHP flaw sent 121 companies into emergency patching
On October 21 the PHP project shipped emergency fixes for two critical flaws, nicknamed PHP2Shell (CVE-2026-109618 and CVE-2026-109619). PHP is the programming language behind a large share of the world's websites. The flaws let an attacker take over a web server without logging in. Details leaked before the fixes were ready, and Amazon said one of its decoy servers saw an attack attempt on October 17, four days before the public announcement. The fixed versions are 8.2.35, 8.3.36, 8.4.27 and 8.5.12. Our own data shows how wide this went: from October 21, 121 of the companies whose status pages we read posted PHP security notices or emergency maintenance, 142 notices in all. Web hosts, software firms and education platforms are all on that list. SquidGrid itself runs no PHP and was not affected.
Help Net Security: PHP2Shell vulnerabilities patched, fixed versions, leak and Amazon honeypot (Oct 21) · BleepingComputer: PHP ships emergency patches for critical PHP2Shell flaws
2026-10-20 Google Cloud's London region lost a data center to a power failure
For about six hours on October 20, from 06:00 to 12:10 UTC, dozens of Google Cloud services in London (the region Google calls europe-west2) had problems, including its servers, databases and storage. Google's own report says the power failed at the building that houses one of the region's three zones, and a fault stopped the backup generators from taking over, so the servers there went down. Google calls this a preliminary finding; its full review is still to come. Some online reports said UK banks and public services were hit; we could not confirm those claims from a primary source, so we are not repeating them.
Google Cloud status page: europe-west2 multi-service incident, preliminary root cause (power disruption, generator start fault)
2026-10-24 Crimea went dark after a strike on its power supply
Overnight into October 24, an attack damaged the Taman electrical substation in Russia's Krasnodar region, which feeds power to occupied Crimea. Sevastopol and Yalta lost electricity, and residents later reported new blackouts in Simferopol, Yalta and Alushta. The internet monitor NetBlocks reported widespread internet outages across Crimea, which follow when the power goes. Local officials said power came back in several cities, but no full restoration time was given. Our routes do not measure inside Crimea separately, so we cannot add our own numbers here.
Kyiv Independent: Mass blackouts reported across occupied Crimea (Oct 24), citing NetBlocks

Ongoing, week 19 of tracking: Iran, better than last week. Routes into Iran (57 we measure) ran 0.3 milliseconds slower than normal, down from 0.9 last week. On October 21 the Iranian economic news site Zoomon reported that unnamed higher authorities had ordered international internet access to be cut immediately if Iran is attacked again. Zoomon did not publish the order or say who issued it, and the authorities have not confirmed it. Earlier, on October 10, deputy communications minister Vahid Yazdanian said the decision was not his to make and that restrictions were possible. Officials give differing accounts of who controls such a shutdown.

We cover earthquakes of magnitude 7.0 or more and storms at the top two disaster alert levels. “Our fleet’s view” is what our own measurements saw, not a damage report.

Subsea watchwhere the internet is wet

China’s internet gateways

China’s three big carriers control most traffic in and out of the country. We check how much of each network we can reach from outside China. The percentage is how much we reached; “normal” is its usual level. Lower means more is blocked or unreachable.

China Telecom 57% reachable (normal 54.5%)
About normal
China Mobile 91.5% reachable (normal 89.3%)
About normal
China Unicom 73.3% reachable (normal 75.7%)
About normal

“About normal” means within 3 percentage points of the usual level.

For the feedpaste-ready

The Squid's Ink -- Week 44 This week's top story: a critical flaw in PHP, the language behind a large share of the web, sent companies scrambling. The PHP project shipped emergency fixes on October 21 for two flaws, nicknamed PHP2Shell, that let an attacker take over a web server without logging in. Details leaked before the fixes were ready. From the status pages we read, 121 companies posted security notices or emergency maintenance within days. Also this week: a power failure, and backup generators that did not start, took down part of Google Cloud's London region for about six hours on October 20. A strike on a substation left occupied Crimea without power and largely offline on October 24. And in Iran, a news site reported an unconfirmed order to cut international internet access immediately if the country is attacked again. Full report in The Squid's Ink. #InternetHealth #CyberSecurity #CloudOutage

Every figure derives from SquidGrid’s own measurement fleet and its archive of provider-published incidents; provider self-reports are labeled as such. Provider Assessment Reports carry the per-provider detail. The internet was mostly fine this week. We measured anyway.