Dive Brief:
- Network delivery service provider Cloudflare suffered "network performance issues" Tuesday morning, according to the company's incident report page. About an hour after the incident began, Cloudflare confirmed it was completely resolved and operations were restored.
- Cloudflare's CEO Matthew Prince took to Twitter saying there was a "massive spike in CPU usage," resulting in primary and backup systems to collapse. The company is still investigating the "root cause" though there is no evidence of malicious activity yet.
- Cloudflare customers experienced 502 gateway errors while trying to access sites served by Cloudflare. The company is continuing to investigate the incident, according to Cloudflare's system status. Impacted regions span the globe. Customers in select cities in Serbia, India, the Philippines, Brazil, Iraq, Saudi Arabia, the U.S. and Australia have been rerouted.
Dive Insight:
The web runs in parts but is ultimately intertwined.
Last month the internet, including Cloudflare, experienced a "massive route leak" that affected most of the internet. The company described the event as "a small heart attack."
The incident was caused after a small Pennsylvania company "became a preferred path of many internet routes through Verizon," which should have never occurred, according to Cloudflare.
Verizon's actions ultimately sent shockwaves through the rest of the web. Because the internet is comprised of a network of networks, it creates an intimate interconnectedness. When one provider goes down, a domino effect begins.
Cloudflare's customers, including Glossier, HubSpot and AutoTrader, use the company's services to protect its sites from DDoS attacks, malicious bots and other cyber activities. The company has yet to state the cause of Tuesday's outage.
In less than a week, the internet has seen two companies impacted by outages.
Last Friday, Slack experienced "degraded services" that lasted more than 12 hours. The issue was fully resolved by 10:20 p.m. EST that day. Slack's outage began when some its servers became unavailable, but then a "separate issue increased server load and dropped a large number of user connections," according to the company's status page.
