I think every team that runs internet-facing web servers should hold itself to a seven-day clock for security updates, starting the day a vendor ships them, and should write that number down before the next disclosure forces the argument.
The case rests on three disclosures that arrived close together. Wiz Research groups PHP2Shell, Java4Shell and the NGINX flaws fixed on November 18 as memory-safety bugs that were historically extremely difficult to discover and exploit. Its researchers found the two HTTP/2 flaws in NGINX with an AI scanning tool, and five separate research teams reached them independently. Wiz puts the conclusion plainly: “Complexity can no longer be relied upon as a barrier to discovery or exploitation.”
That sentence matters for how a patch policy is written. Most policies I have read set windows by severity score: critical in 14 days, high in 30. Those numbers were chosen when finding a bug like this took a specialist months. The policy still assumes the gap between a vendor fix and a working attack is long, and the published evidence points the other way. Rapid7 built a working demonstration against an NGINX 1.31.8 container and published it with its own analysis on the day of disclosure. The embargo ran about four weeks from Wiz’s first report on October 21, and Wiz notes that F5 supplied a patch for review on November 10. The vendors had the fix well before most customers saw it.
The strongest objection
Operations teams will say a seven-day clock on the web tier invites outages, and that the NGINX flaw is harder to use than the headline suggests. They have a point on the second half. Rapid7 reports that its demonstration needs about 128 GB of HTTP/2 data, which takes seventeen minutes on a 1 Gbit/s link, and that NGINX’s default one-hour connection limit stops attackers on networks slower than about 300 Mbit/s. CIRCL adds that code running as a worker does not by itself give root on the host.
I accept all of that and it still does not change the answer. The same Rapid7 numbers show two minutes on a 10 Gbit/s link, and an attacker on a data center link has that. CIRCL’s guidance is that “neither ASLR nor PIE should be considered a substitute for patching,” which removes the usual reason to downgrade a ticket. On the outage worry, Wiz itself says turning a published fix into protection takes substantial manual work, from validating patches to coordinating rollouts with service owners. That work is the reason to set a deadline. A seven-day clock forces the staging build to exist, and a reverse proxy tier that cannot take a tested build within a week is the finding that deserves the budget conversation.
What the clock requires
A deadline only works if three things exist before the advisory lands. The first is an inventory that says where NGINX, PHP and Java runtimes run, including the copies inside container images and ingress controllers, because a WAF rule bought time on PHP2Shell without removing the flaw. The second is a configuration record. CIRCL’s table shows that turning off HTTP/2 blocks two of the five NGINX paths and leaves three open, so a team has to know which features each host enables before it can judge a stopgap. The third is an owner with the authority to restart a production tier, named in advance.
A seven-day clock also gives teams a fair test of their own pipelines. The Java4Shell fixes landed the day before the NGINX ones. A team that cleared the first within a week had practiced for the second.
The deadline also changes who has to say yes. With a written number, a delay becomes a risk the service owner signs for, with a name and a date, instead of an unexamined default. I would rather see ten of those signatures on file than a policy nobody has read.
My recommendation is concrete. Pick the number, publish it to the business, and measure it monthly as the share of internet-facing hosts running a build less than seven days behind the vendor’s latest fix. Start with the web server and proxy tier, since those hosts take traffic from everyone. If the share is under 90 percent after two months, treat that as a funding request for staging capacity.
Source: Wiz Research