Lithium battery fire at Daejeon Data Center in South Korea

Comments

• david Member

  October 2025 edited October 2025

  Latest word is that 858 TB were lost in the fire. They had some backups, but apparently some only monthly.

  The National Information Resources Service (NIRS) in South Korea, however, recently lost 858TB of government data after a fire broke out in the building – and there were no backups.

  According to the National Information Resources Service, of the 647 systems at the Daejeon headquarters, 62% were backed up daily, but the remaining 38% were backed up once a month. Some systems were last backed up on August 31.

  https://www.techspot.com/news/109799-858tb-gone-forever-south-korea-data-center-fire.html

  Thanked by 1oloke
• Calypso Member

  October 2025

  @david said: According to the National Information Resources Service, of the 647 systems at the Daejeon headquarters, 62% were backed up daily, but the remaining 38% were backed up once a month. Some systems were last backed up on August 31.

  Well that's not necessarily has to be an issue; if those 38% are all frontend servers without data on it, less frequent backup doesn't need to be an issue - even no backups can do in that case when you have an automated deployment in place. Quite often even faster as a restore.
• Rubben Member

  October 2025

  it was @nekomikoreimu

  Thanked by 1nekomikoreimu
• jnd Member

  October 2025

  This is why you need to have offsite backups.
• jnd Member

  October 2025 edited October 2025

  @Calypso said:

  @david said: According to the National Information Resources Service, of the 647 systems at the Daejeon headquarters, 62% were backed up daily, but the remaining 38% were backed up once a month. Some systems were last backed up on August 31.

  Well that's not necessarily has to be an issue; if those 38% are all frontend servers without data on it, less frequent backup doesn't need to be an issue - even no backups can do in that case when you have an automated deployment in place. Quite often even faster as a restore.

  If you can rebuild the services from source then it's fine. But 858 TB doesn't sound like frontend servers. Unless they all use node.js.

  From the article:

  G-Drive, which stands for Government drive rather than Google, stored data from around 125,000 federal workers – who were allowed 30GB each – and was core to 163 public-facing services, holding everything from import and export certifications to product safety checks, according to reports.
  
  What's stunned South Korea is the revelation that the G-Drive was not backed up. An official told The Chosun that the drive could not be backed up due to its large size. About 858TB of data was lost. Ninety-five other systems in the fire were also destroyed, but they were backed up.
• emgh Member, Megathread Squad

  October 2025

  @jnd said: An official told The Chosun that the drive could not be backed up due to its large size.

  lmfao

  Thanked by 1jnd
• bbn12 Member

  October 2025

  who needs backups...
• david Member

  October 2025

  It sounds like a job for Host-C. Just spin up an extra 900 TB.

  Thanked by 2Rubben jnd
• Calypso Member

  October 2025

  @jnd said:

  @Calypso said:

  @david said: According to the National Information Resources Service, of the 647 systems at the Daejeon headquarters, 62% were backed up daily, but the remaining 38% were backed up once a month. Some systems were last backed up on August 31.

  Well that's not necessarily has to be an issue; if those 38% are all frontend servers without data on it, less frequent backup doesn't need to be an issue - even no backups can do in that case when you have an automated deployment in place. Quite often even faster as a restore.

  If you can rebuild the services from source then it's fine. But 858 TB doesn't sound like frontend servers. Unless they all use node.js.

  I was mainly looking at the 62/38% - that it doesn't need to be an issue. The 858T of "lost data" ofcourse is.