Causes Behind The Major Disruption To The System At ICA Checkpoints On 31st March 2023
Several hon Members and I filed parliamentary questions seeking an explanation on the 5 hour system outage at ICA checkpoints on 31 March 2023 that led to major inconveniences to members of public. In his response, Minister of State Assoc Prof Dr Muhd Faishal Ibrahim informed that the system crashed when ICA was conducting a trial to upgrade the system to allow for immigration clearance using biometrics. He also outlined some plans for the future including the next generation automated border clearance systems. My PQ and MOS Faishal’s answer may be accessed below.
— — — — — — — — — — — — — — — — — — — — — —
Mr Murali Pillai asked the Minister for Home Affairs (a) what is the estimated number of people who were affected by the five-hour system disruption at the land and air checkpoints on 31 March 2023 arising from a scheduled system upgrade that caused a system overload; (b) why was the problem hitherto unanticipated; © what is the Ministry’s assessment of the contingency plan activated to deal with the disruption; and (d) what steps will be taken to strengthen the plan.
Assoc Prof Dr Muhammad Faishal Ibrahim:
[…] On 31 March 2023, at about 10.40 am, the Immigration and Checkpoints Authority (ICA) was conducting a pre-scheduled trial needed for an upgrade of the Multi-Modal Biometrics System (MMBS). The MMBS facilitates automated immigration clearance using travellers’ biometrics.
Unfortunately, the trial caused a severe system overload. Automated clearance lanes at all the departure halls in Changi Airport, and certain automated lanes at Woodlands and Tuas Checkpoints were impacted. The rest of the automated lanes, including those at the sea checkpoints, remained operational.
Prior to the incident, ICA had been upgrading its systems progressively and cautiously, with 10 out of 12 systems enhanced. Only MMBS and one other system had not yet been upgraded. For MMBS, the system upgrade would involve replicating the large biometric database to an upgraded system. This has to be done continuously over a few days and the MMBS has to remain operational during this period.
Our standard procedure for any system upgrade is to conduct tests in the User Acceptance Test (UAT) environment before deployment. The UAT tests were successful. Subsequent trials in the production environment were also stable.
Next, we had planned to conduct further controlled trials during different times of the day, to ascertain that the system upgrade would not disrupt operations during the wee hours, off-peak hours and peak hours, before proceeding with the actual upgrade.
The wee hours trial was conducted on 15 March 2023, from 1.30 am to 3.30 am. The trial was successful.
The trial on 31 March 2023, when the incident happened, was the off-peak hours trial, to take place from 10.00 am to 2.00 pm. The vendors were on standby on-site, and the plan was to recover the system within 30 minutes if the trial did not go well.
The trial caused the storage systems to overload at about 10.40 am and the process was aborted immediately. However, the extent of the overload was much more severe than anticipated and the vendors who were onsite had to work with their global support team to diagnose and reboot the servers. In the end, the recovery process took about four and a half hours and MMBS recovered around 3.00 pm.
ICA immediately activated its Business Continuity Plan (BCP) once the system went down. Off-duty officers were recalled to help man manual immigration counters and perform crowd control. Across all the checkpoints, the failover process kicked in, and all the manual counters and certain automated lanes, switched to back-up systems. Not all the automated lanes have this failover capability, as different models were procured over the years.
At the land checkpoints, travellers experienced delays of up to 30 minutes at the onset of the incident. At Woodlands Checkpoint, there was later another hour of delay for cars as the car arrival zone had to be converted to clear motorbikes manually. About 55,000 travellers passed through the two land checkpoints during the disruption.
Immigration clearance at the Airport’s departure halls was significantly impacted. Travellers were re-directed to manual counters for immigration clearance. ICA immediately stepped up manning of the manual counters through a combination of measures, such as recalling off-duty officers, deploying administrative staff and retaining the officers from the outgoing shift.
Changi Airport Group (CAG) assisted ICA by deploying additional Changi Youth Ambassadors and office staff, along with CAG’s Duty Terminal Managers and their Customs, Immigration and Quarantine Team. They helped identify travellers whose flights were departing soon, so that their departure clearance could be prioritised. Public announcements were also made to appeal for such travellers to step forward.
During the disruption, about 30,000 passengers on 113 departure flights and 111 arrival flights, cleared immigration. Despite the measures taken by ICA and CAG, 21 out of the 30,000 cleared passengers missed their flights at Terminal 4. These travellers were subsequently offered to re-book on alternative flights within a week at no extra cost.
In light of the incident, ICA is reviewing the approach to the upgrade. The alternative approach that ICA was initially offered by the vendors was not tenable. It involved upgrading the system directly without replicating the database but may require five days of system downtime.
Even with the implementation of BCPs, degradation in service standards was inevitable, given the volume of travellers, and especially, at the onset of the incident. We apologise to affected travellers for the inconvenience caused and thank them for their understanding. I would also like to thank the ICA officers and CAG staff who had worked tirelessly, without complaint, to deal with the situation.
Going forward, ICA is in the process of implementing a multi-year plan, known as the New Clearance Concept (NCC), to transform immigration clearance across all the checkpoints. NCC will provide more seamless clearance and help ICA, with limited manpower resources, cope with the ever-rising traveller volume. Under NCC, all manual counters and automated lanes will be replaced in phases with the next generation Automated Border Clearance System (ABCS) gates, which will allow travellers to continue using automated clearance, even when MMBS is down.
In the meantime, the resiliency of existing automated lanes that have yet to be replaced by the ABCS gates will also be enhanced in the second half of 2023.
Sir, Singapore is one of the first countries in the world to implement NCC-type capabilities. Some of these capabilities will be implemented in a form and manner not used or tested elsewhere, as we customise them to suit our operational needs. As such, while we will work closely with all stakeholders to further minimise technical disruptions and enhance system resiliency, the reality is, disruptions will still happen now and then. When they do, we will make sure we have robust contingency plans in place and try to recover as fast as possible. I hope for the public and Members’ understanding.