On September 27, operations of 647 government work systems were suspended due to a fire in the computer room at the National Information Resources Service (NIRS) in Daejeon, causing significant inconvenience. There are growing criticisms that insufficient redundancy in the cloud environment exacerbated the situation.
NIRS Fire Reveals Lack of Cloud Redundancy
If there had been a 'twin' cloud system with the same environment, it could have functioned in another region in the event of a disaster at one site, thereby preventing a large-scale outage. However, it has been revealed that such a service redundancy (backup) system is not properly in place at present.
At a briefing held at the Government Seoul Office that day, Vice Minister of the Ministry of the Interior and Safety Kim Minjae stated, "This fire broke out in one battery, whose power had been cut off, during work to relocate the NIRS uninterruptible power supply battery underground," and added, "We will prioritize restoring major government services with significant public impact, such as Korea Post financial and postal services, as quickly as possible."
Jaeyong Lee, Director of the National Information Resources Service (right), is answering reporters' questions at the 'National Information Resources Service Government Service Disruption Briefing' held on the morning of the 27th at the briefing room of the Government Seoul Office in Jongno-gu, Seoul. Photo by Yonhap News
원본보기 아이콘"If There Had Been a 'Twin Cloud'... Large-Scale Outages Could Have Been Prevented"
The computer room where the fire broke out the previous day is known as the 'G-Cloud Zone,' a private cloud environment operated by the National Information Resources Service. The disaster recovery (DR) system for this area requires both server DR and cloud DR. However, it has been analyzed that while the NIRS had a disaster recovery environment for servers, the disaster recovery environment for the cloud was not yet completed, which contributed to the complete shutdown of government systems and escalation of the incident.
Deja Vu of the Kakao Incident Three Years Ago
The paralysis of government systems caused by the NIRS fire overnight has drawn comparisons to the Kakao service outage three years ago. In October 2022, a fire at the data center housing KakaoTalk servers led to simultaneous disruptions across most Kakao services, including the national messenger KakaoTalk. Although some services, such as KakaoTalk, were restored after overnight recovery efforts, many users experienced significant inconvenience and confusion.
Following the outage, Kakao announced plans to upgrade its disaster recovery system to a triple-redundant system linking three data centers and promised to enhance service stability through infrastructure improvements. The Kakao outage highlighted the importance of data center management, but the recent NIRS fire has once again exposed shortcomings in management.
An official from the cloud industry stated, "Given the importance of these resources, the disaster recovery (DR) system for the affected area requires both server DR and cloud DR. While the fire itself was the primary cause, the lack of a proper cloud DR system led to a complete government system shutdown and widespread disruption."
Delays Due to Budget Issues... Redundancy Work Not Completed
Meanwhile, it has been reported that although there were plans to establish redundancy between the NIRS Daejeon headquarters and the Gongju center, progress was delayed due to budget issues. Earlier this year, a detailed plan for building the cloud disaster recovery system was proposed, with a sequential migration scheduled over five years and detailed consulting to begin next year. However, there have also been objections raised regarding the transfer of government cloud systems to the private sector during this process.