Several major websites and services, including Amazon, Steam, and Airbnb, were down for about an hour on Thursday due to issues with Akamai's Edge DNS service
Update 12:48 p.m. ET: Akamai says it has implemented a fix for the issue. “We have implemented a fix for this issue …
Context & Ripple Effects
The outage follows a [[a:967161|June Fastly disruption that simultaneously affected Amazon, Twitch, Spotify, Reddit and major publishers]], showing that a fault at a shared web-delivery provider can surface as failures across otherwise unrelated consumer services.
It also echoes the Dyn DNS attack that disrupted popular US East Coast sites, but Akamai says its Edge DNS service was restored after about an hour. The immediate issue is not one company’s application availability, but a common layer used to reach multiple services.
First-order effects
- Amazon, Steam, Airbnb and other affected sites lost availability for roughly an hour; Akamai’s implemented fix restored the Edge DNS layer serving them.
- Akamai’s customers must treat the incident as an operational dependency failure even though the underlying applications were not identified as the source of the outage.
Second-order effects
- The back-to-back Fastly and Akamai incidents put pressure on large online services to scrutinize DNS and delivery-provider concentration, including whether a single provider can interrupt customer access across their properties.
- Competing edge and DNS providers gain a concrete reliability comparison point, while affected businesses face added demand for outage communications and recovery planning.
Third-order effects
- Repeated outages spanning a CDN, a DNS provider and Amazon’s own cloud operations point to web resilience becoming a supply-chain problem: service availability increasingly depends on a small set of shared infrastructure layers.
- If operators respond by diversifying those layers, edge and DNS vendors will compete not only on performance and price but also on customers’ ability to limit the blast radius of a provider failure.
The trend: Consumer internet services are becoming more exposed to failures at concentrated infrastructure layers, pushing resilience and dependency diversification higher in platform buying decisions.