Impact of the Google outage can be measured in GDP
August 20 2020
By David Richards
Google’s five-hour disruption was concentrated in the UK, Greece, Spain, Germany, France, Japan and Malaysia
If the internet is our information superhighway, Thursday's mass outage of Google services represents the sudden and total closure of the M25. Users up and down the country who rely on the system for their livelihoods found themselves confronted with the simple Gmail message: “Oops, something went wrong”. It was the digital equivalent of the Road Closed sign.
Such is the public and private sector’s dependence on software services provided by Google and its rivals Amazon, Microsoft and Alibaba that the five-hour outage will likely be felt at GDP level. Never mind the frustration felt by hundreds of thousands of homeworkers, think of all the lost opportunities from meetings unattended, the lost confidence from work unsent and the lost productivity from reduced output.
It all adds up: a temporary internet shutdown costs an advanced economy like Britain’s $141m (£107m) per day, according to a report from Deloitte and Facebook into the economic impact of disruptions to connectivity. That’s equivalent to 1.9pc of daily GDP. A big hit, especially in a recession when companies small and large are fighting for their lives and public services are stretched to the limit.
As a provider of big data software, I’m sometimes asked if the UK’s reliance on so-called cloud computing is a weakness. Actually, having access to enormous computing power at a fraction of the cost is a strength. It enables businesses and organisations to transform themselves for the digital age and scale up their products and services. Without cloud computing, it would be impossible to work at home during the pandemic. Then our economy really would have fallen off a cliff.
The problem here is with resilience. Every computing system has outages and they happen all the time. Some are planned, usually for weekends, and some unplanned, usually by accident. The issue is how long a system can be down - Recovery Time Objective - and how much data can be lost - Recovery Point Objective - without significant damage to the application, not to mention reputation. Take British Airways and its recent history of IT failures: the flag carrier has suffered a series of outages causing chaos and cancellations for customers.
When we dealt in small quantities of data, outages didn’t matter so much. In the new world, we are generating unimaginably large amounts of data through widespread use of technology in every aspect of our lives. Unfortunately, big data also means big outages. When cloud providers say they can guarantee against outages 99.99 per cent of the time, it might sound impressive but the cost of marginal error becomes substantial. Big tech firms could, and should, be guaranteeing durability of at least 99.99999 per cent.
The solution to this problem of leaky plumbing is what we call geo redundancy. This means the ability of applications to “fail over” to exact replicas of data stored in entirely different physical locations. My company, WANdisco, is providing these services to international airlines, supermarkets and healthcare providers as they move core operational infrastructure to the cloud.Read more