That's one disaster recovery -- and planning -- takeaway from the latest "Annual Incident Reports" study, published Tuesday by the European Network and Information Security Agency (ENISA).
The report is based on 79 incident reports of severe outages of electronic communication networks or services that occurred in 2012. ENISA defines serious incidents -- which European government agencies are required to report -- on a sliding scale, ranging from events that last at least an hour and affect 15% or more users, up to events that last more than eight hours and affect 1% or more of all users.
All told, 18 European countries reported having a total of 79 "significant incidents" to ENISA, while nine other countries said they had no such incidents.
[ Twitter users should reassign access to third-party apps. Read more: Hacker Leaks 15,000 Twitter Access Credentials, Promises More. ]
Downtime is a major concern for businesses and government agencies. When systems are down, employee productivity plummets, sales get put on hold, subscribers can't be billed and emergency services can't be reached.
The report assessed four different types of service interruptions: fixed telephony and fixed Internet, and mobile telephony and mobile Internet. For incidents in all four services, hardware failure was the most common cause of service disruption, according to the report. "The second most common cause for fixed telephony was [a] software bug," it read. "Half of those incidents affected VoIP." For fixed Internet outages, cyber attack was the second most common cause, reported ENISA. Finally, software bugs were the second most common cause of outages for mobile telephony and mobile Internet.
But the outages with the longest recovery time were triggered by incidents caused by natural phenomena -- mainly storms and heavy snowfall, reported ENISA. On average, such outages lasted about 36 hours.
The majority of outages -- comprising about half of all reports -- involved mobile telephony or mobile Internet and affected an average of 1.8 million people in every incident.
In 75% of outages, the leading root cause was listed as "system failure." "Hardware failures were the most common cause, followed by software bugs," reported ENISA. "The assets most often affected by system failures were switches -- e.g., routers and local exchange points -- and home location registers." The latter refers to databases of authorized mobile phone subscribers.
The report detailed a number of actual incidents, ranging from the natural and the accidental to human error and malicious insiders. As an example on the human-error front, one telephony employee committed a system configuration error that prevented landline users from making outgoing international calls to Western Europe for four hours, which was solved by reconfiguration and a reboot of the system involved. Meanwhile, a 14-hour outage involving 40,000 VoIP users was triggered by the service provider transitioning to a new system and finding it couldn't handle the volume of related registration requests being issued by subscribers' devices.
On the accident front, four submarine cables linking two islands being accidentally cut by a ship's anchor, leading to a temporary outage.
On the malicious front, a distributed denial of service (DDoS) attack against a mobile Internet provider prevented 2.5 million users from connecting to the Internet for up to two hours. After mitigating the attack, the provider put stronger DDoS defenses in place to withstand repeat attacks.
Finally, a former employee of an Internet service provider deliberately set fire to a switching system, which was used for providing fixed Internet service to around 10,000 subscribers, according to ENISA. Service was restored some 36 hours later.