When ‘the cloud’ hits a storm
Surprising events can highlight problems with technology. Tom Pullar-Strecker reports.
In an increasingly interconnected world it is no longer safe to be completely smug about bad weather in other countries.
The downside of cloud computing was exposed last Sunday when weekend storms in Sydney temporarily prevented Kiwis watching internet television service Lightbox, or ordering pizzas from Domino’s.
Ironically, Lightbox chief executive Kym Niblock was doubly affected as she was trying to order pizzas from Domino’s for her children and babysitter when she got a text message saying Lightbox was down.
She tried Pizza Hut but that was slow because of the extra online traffic. ‘‘I gave up and went to a pizza place at the end of the road.’’
The cause of the problem was that Lightbox and Domino’s website are among a raft of services run out of computer centres in Sydney, operated by Amazon Web Services (AWS), a subsidiary of United States retail giant Amazon.
But both went offline for about an hour on Sunday evening after an electricity substation in Sydney was knocked out by the severe storm.
Some other Kiwi businesses that use AWS’ services, including Westpac New Zealand, Xero and Loyalty NZ, reported no major issues.
Don Christie, director of Wellington technology firm Catalyst IT, said the incident should be a wake-up call. Catalyst is a rival to AWS but also helps some of its clients use AWS.
In years gone by, when companies tended to host their own computer system, natural disasters in one country would not have affected businesses in another.
But many Kiwi companies have been moving their information technology systems to AWS because of its scale and the availability of a fibre-optic cable to carry internet data under the Tasman Sea. Christie said that meant the economic impact of outages could be massive.
‘‘No matter how good AWS are, and they are good, that is something that we as a country and the Government need to be thinking about.
‘‘It is important to make sure the whole country is not entirely reliant on one homogenous supplier.’’
Christie also criticised the five days it took AWS to provide a full explanation of what went wrong in Sydney.
‘‘We were putting out reports to our clients within hours,’’ he said.
AWS should have been able to keep all its systems running despite the failure of the substation.
But it said in its incident report released on Thursday, that the power back-up facilities in one of its data centres didn’t kick into action as planned.
Companies such as AWS invest millions of dollars in specialist power conditioning and battery systems called ‘‘uninterruptible power supplies’’ (UPSs) and diesel generators to keep data centres online in case of power failure.
But their performance doesn’t always live up to expectations.
Last Sunday, the issue was that the substation failure caused a long drop in mains voltage, instead of a complete power outage, and that failed to trigger some of AWS’ UPS systems to disengage from the power grid.
The result was their stored energy quickly drained completely into the grid, leaving them unable to switch to their integrated diesel generators which had by then whirred into action.
Niblock said: ‘‘Amazon is a very professional business and we are always happy to deal with them,’’ she said.
‘‘It’s a shame something like this happened; it is very unusual for a business like them.’’
Generally, however, IT systems are getting more reliable thanks to cloud computing.
Two US analysts, Oppenheimer and Bernstein, are respectively forecasting Amazon shares will climb from their current price of US$726 ($1022) to US$930 or US$1000 within the coming year on the back of AWS’ growth.
This would take Amazon’s market value to more than US$430 billion.
So for Amazon, at least, the effects of Sydney’s bad weather may prove a storm in a teacup.