THE voltage fluctuation, caused by lightning, lasted barely 0.1 second, but this seemingly minor occurrence ended up paralysing trading on the Singapore Exchange (SGX) for more than three hours on Wednesday.
After the power fluctuation, SGX's backup - the uninterruptible power supply (UPS) systems - should have kicked in but malfunctioned, leaving its own officials puzzled.
The regulator is not amused. A spokesman for the Monetary Authority of Singapore (MAS) said yesterday that it takes a "serious view of the trading halt".
"We have instructed SGX to take immediate remedial action to prevent a recurrence. MAS has also instructed SGX to conduct a thorough investigation to find out the root cause of the outage and submit its investigation findings to MAS. MAS will not hesitate to take appropriate regulatory and supervisory actions against SGX should lapses be uncovered in the investigations."
An SP PowerGrid spokesman said there was no disruption in electricity supply to its customers. In fact, most of them were not affected by the voltage fluctuation at all.
"For the handful of customers with sensitive equipment, and hence more susceptible, it is industry practice to have protection systems against voltage fluctuations," the spokesman said.
But both SGX's protection systems to its main data centre malfunctioned at 2.18 pm on Wednesday, halting trading for three hours. Trading was back to normal yesterday.
SGX president Muthukrishnan Ramaswami said at a briefing yesterday that this was the first time its UPS systems, designed to provide four layers of protection against supply disruptions, malfunctioned. "It sounds ridiculous to have a backup of a backup of a backup that fails... It's not normal," he said. "UPS systems are designed to withstand power fluctuations, which are normal. But the component that was supposed to withstand it failed."
He added: "We have yet to get to the root cause of why the UPS malfunction happened."
SGX said fluctuation in power supply affected its primary data centre. Then the backup internal power source also malfunctioned. It was reluctant to rely solely on the secondary data centre, in case it was not properly synchronised with the primary one. It chose, instead, to reboot both trading engines.
"If we had this (problem) again, then we will ensure the data centre that goes down will stay dead, and the switch-over to the secondary data centre will happen quicker in terms of recovery," said Mr Ramaswami.