Web hosting firm hostinguk suffered a major outage at their St Asaph data centre. Despite having “diverse fibre paths” to the site, the connection to the site was lost just before 9pm last night. For many customers, the shock of finding that their website was down was compounded by the fact that they couldn’t contact hostinguk because their phones were not working.
Credit where credit is due. Unlike may firms in the midst of an incident, hostinguk communicated well throughout the disruption; the status page on their website was regularly updated with detailed information on the progress of the recovery, and social media posts were responded to promptly. However, there are two very interesting aspects of this incident.
Firstly, the initial estimate for “full service restoration” was “2-12 hours”; in actual fact, service was restored by way of a temporary fix after 17 hours and the permanent repair took a further four hours. This is an excellent example of the prevalence of “optimistic overconfidence”: people tend to be overly optimistic about favourable outcomes and discount the probability of less positive scenarios. This phenomenon is often then exacerbated by “anchoring”: it becomes clear over time that the initial estimates are wrong but people are still “anchored” to them and fail to adjust their estimates sufficiently.
Secondly, what does “diverse fibre paths” actually mean? The reality is that a single event managed to sever both fibres so I would caution against being too reassured by this statement from your providers.